Add public MethodInvoker and ConstructorInvoker classes

[steveharter]

Fixes #85539

Adds fixed-parameter-count and Span-based Invoke APIs for performance, primarily achieved by preventing the object[] allocation for when there are <= 4 parameters. Also, additional, faster, emit-based functions are generated for methods that don't have ref\out parameters and meet other constraints. These are also used by the existing Invoke methods so performance there has increased as well.

Todo

• Add doc for the new public methods
• Create PR for Mono to address unboxing a "true nullable" (currently the PR doesn't generate faster emit-based code for Mono). Done - see Enable Mono for using faster invoke stubs #89108
• Add tests that layer on top of existing ones; currently this PR just adds a handful of new tests for the new APIs.
• Add the new benchmarks to the performance repo

Benchmarks

Benchmarks for the new APIs. ~2.4- 3.3x faster for the 4 parameter case with no allocs. Also, these improvements would be greater but they are compared to the existing Invoke API which was also made faster here.

Method	Mean	Error	StdDev	Median	Min	Max	Gen0	Allocated
Method0_NoParms	7.230 ns	0.0205 ns	0.0181 ns	7.228 ns	7.199 ns	7.265 ns	-	-
Method0_NoParms_InvokerEquivalent	6.068 ns	0.0365 ns	0.0342 ns	6.066 ns	5.994 ns	6.123 ns	-	-
StaticMethod4_arrayNotCached_int_string_struct_class	41.669 ns	0.2680 ns	0.2238 ns	41.627 ns	41.169 ns	41.979 ns	0.0098	104 B
StaticMethod4_int_string_struct_class_InvokerArgsEquivalent	17.156 ns	0.1327 ns	0.1242 ns	17.164 ns	17.004 ns	17.391 ns	0.0045	48 B
StaticMethod4_arrayNotCached_PreBoxed_int_string_struct_class	37.078 ns	0.2929 ns	0.2446 ns	37.064 ns	36.747 ns	37.647 ns	0.0053	56 B
StaticMethod4_arrayNotCached_PreBoxed_int_string_struct_class_InvokerArgsEquivalent	11.160 ns	0.0615 ns	0.0576 ns	11.168 ns	11.057 ns	11.244 ns	-	-

Benchmarks for existing Invoke APIs. Up to 1.4x faster for a property setter. Fewer allocs when >=5 parameters.

|                                                    Method |        Job |                  Toolchain |       Mean |     Error |    StdDev |     Median |        Min |        Max | Ratio | RatioSD |   Gen0 | Allocated | Alloc Ratio |
|---------------------------------------------------------- |----------- |--------------------------- |-----------:|----------:|----------:|-----------:|-----------:|-----------:|------:|--------:|-------:|----------:|------------:|
|                                           Method0_NoParms | Job-MZSTNH |  \INVOKE_AFTER\corerun.exe |   7.625 ns | 0.0528 ns | 0.0441 ns |   7.624 ns |   7.549 ns |   7.712 ns |  1.00 |    0.00 |      - |         - |          NA |
|                                           Method0_NoParms | Job-UYFZWZ | \INVOKE_BEFORE\corerun.exe |   9.401 ns | 0.0397 ns | 0.0352 ns |   9.404 ns |   9.327 ns |   9.458 ns |  1.23 |    0.01 |      - |         - |          NA |
|                                                           |            |                            |            |           |           |            |            |            |       |         |        |           |             |
|      StaticMethod4_arrayNotCached_int_string_struct_class | Job-MZSTNH |  \INVOKE_AFTER\corerun.exe |  40.351 ns | 0.5224 ns | 0.4631 ns |  40.182 ns |  39.786 ns |  41.220 ns |  1.00 |    0.00 | 0.0099 |     104 B |        1.00 |
|      StaticMethod4_arrayNotCached_int_string_struct_class | Job-UYFZWZ | \INVOKE_BEFORE\corerun.exe |  49.067 ns | 0.9783 ns | 1.0468 ns |  48.853 ns |  47.515 ns |  51.704 ns |  1.22 |    0.03 | 0.0098 |     104 B |        1.00 |
|                                                           |            |                            |            |           |           |            |            |            |       |         |        |           |             |
| StaticMethod5_arrayNotCached_int_string_struct_class_bool | Job-MZSTNH |  \INVOKE_AFTER\corerun.exe |  56.730 ns | 0.5781 ns | 0.5408 ns |  56.669 ns |  55.909 ns |  57.890 ns |  1.00 |    0.00 | 0.0128 |     136 B |        1.00 |
| StaticMethod5_arrayNotCached_int_string_struct_class_bool | Job-UYFZWZ | \INVOKE_BEFORE\corerun.exe |  74.827 ns | 1.6601 ns | 1.8452 ns |  74.496 ns |  71.119 ns |  78.316 ns |  1.32 |    0.03 | 0.0191 |     200 B |        1.47 |
|                                                           |            |                            |            |           |           |            |            |            |       |         |        |           |             |
|                     StaticMethod4_int_string_struct_class | Job-MZSTNH |  \INVOKE_AFTER\corerun.exe |  28.537 ns | 0.2502 ns | 0.2218 ns |  28.585 ns |  28.185 ns |  28.842 ns |  1.00 |    0.00 |      - |         - |          NA |
|                     StaticMethod4_int_string_struct_class | Job-UYFZWZ | \INVOKE_BEFORE\corerun.exe |  34.938 ns | 0.1109 ns | 0.0983 ns |  34.899 ns |  34.821 ns |  35.179 ns |  1.22 |    0.01 |      - |         - |          NA |
|                                                           |            |                            |            |           |           |            |            |            |       |         |        |           |             |
|         StaticMethod4_ByRefParams_int_string_struct_class | Job-MZSTNH |  \INVOKE_AFTER\corerun.exe | 144.779 ns | 1.0213 ns | 0.9054 ns | 144.673 ns | 142.893 ns | 146.116 ns |  1.00 |    0.00 | 0.0041 |      48 B |        1.00 |
|         StaticMethod4_ByRefParams_int_string_struct_class | Job-UYFZWZ | \INVOKE_BEFORE\corerun.exe | 138.456 ns | 0.2052 ns | 0.1714 ns | 138.511 ns | 138.134 ns | 138.783 ns |  0.96 |    0.01 | 0.0043 |      48 B |        1.00 |
|                                                           |            |                            |            |           |           |            |            |            |       |         |        |           |             |
|    StaticMethod5_ByRefParams_int_string_struct_class_bool | Job-MZSTNH |  \INVOKE_AFTER\corerun.exe | 196.652 ns | 1.2706 ns | 1.1885 ns | 196.883 ns | 194.293 ns | 198.425 ns |  1.00 |    0.00 | 0.0063 |      72 B |        1.00 |
|    StaticMethod5_ByRefParams_int_string_struct_class_bool | Job-UYFZWZ | \INVOKE_BEFORE\corerun.exe | 193.105 ns | 1.0413 ns | 0.8696 ns | 193.156 ns | 191.592 ns | 194.599 ns |  0.98 |    0.01 | 0.0123 |     136 B |        1.89 |
|                                                           |            |                            |            |           |           |            |            |            |       |         |        |           |             |
|                                            Ctor0_NoParams | Job-MZSTNH |  \INVOKE_AFTER\corerun.exe |   9.128 ns | 0.0912 ns | 0.0853 ns |   9.137 ns |   9.005 ns |   9.264 ns |  1.00 |    0.00 | 0.0053 |      56 B |        1.00 |
|                                            Ctor0_NoParams | Job-UYFZWZ | \INVOKE_BEFORE\corerun.exe |  11.096 ns | 0.1013 ns | 0.0947 ns |  11.085 ns |  10.965 ns |  11.262 ns |  1.22 |    0.02 | 0.0053 |      56 B |        1.00 |
|                                                           |            |                            |            |           |           |            |            |            |       |         |        |           |             |
|                             Ctor4_int_string_struct_class | Job-MZSTNH |  \INVOKE_AFTER\corerun.exe |  32.506 ns | 0.2559 ns | 0.2137 ns |  32.561 ns |  32.026 ns |  32.865 ns |  1.00 |    0.00 | 0.0053 |      56 B |        1.00 |
|                             Ctor4_int_string_struct_class | Job-UYFZWZ | \INVOKE_BEFORE\corerun.exe |  39.152 ns | 0.2872 ns | 0.2687 ns |  39.151 ns |  38.818 ns |  39.719 ns |  1.20 |    0.01 | 0.0052 |      56 B |        1.00 |
|                                                           |            |                            |            |           |           |            |            |            |       |         |        |           |             |
|                                          Property_Get_int | Job-MZSTNH |  \INVOKE_AFTER\corerun.exe |  10.565 ns | 0.1276 ns | 0.1131 ns |  10.581 ns |  10.416 ns |  10.833 ns |  1.00 |    0.00 | 0.0023 |      24 B |        1.00 |
|                                          Property_Get_int | Job-UYFZWZ | \INVOKE_BEFORE\corerun.exe |  12.435 ns | 0.0953 ns | 0.0845 ns |  12.432 ns |  12.299 ns |  12.604 ns |  1.18 |    0.01 | 0.0023 |      24 B |        1.00 |
|                                                           |            |                            |            |           |           |            |            |            |       |         |        |           |             |
|                                        Property_Get_class | Job-MZSTNH |  \INVOKE_AFTER\corerun.exe |   8.091 ns | 0.0468 ns | 0.0415 ns |   8.093 ns |   8.037 ns |   8.178 ns |  1.00 |    0.00 |      - |         - |          NA |
|                                        Property_Get_class | Job-UYFZWZ | \INVOKE_BEFORE\corerun.exe |  10.549 ns | 0.0877 ns | 0.0778 ns |  10.555 ns |  10.411 ns |  10.685 ns |  1.30 |    0.01 |      - |         - |          NA |
|                                                           |            |                            |            |           |           |            |            |            |       |         |        |           |             |
|                                          Property_Set_int | Job-MZSTNH |  \INVOKE_AFTER\corerun.exe |  16.105 ns | 0.1308 ns | 0.1160 ns |  16.110 ns |  15.916 ns |  16.325 ns |  1.00 |    0.00 | 0.0023 |      24 B |        1.00 |
|                                          Property_Set_int | Job-UYFZWZ | \INVOKE_BEFORE\corerun.exe |  21.829 ns | 0.4245 ns | 0.3971 ns |  21.916 ns |  21.035 ns |  22.320 ns |  1.35 |    0.03 | 0.0022 |      24 B |        1.00 |
|                                                           |            |                            |            |           |           |            |            |            |       |         |        |           |             |
|                                        Property_Set_class | Job-MZSTNH |  \INVOKE_AFTER\corerun.exe |  13.299 ns | 0.0699 ns | 0.0654 ns |  13.304 ns |  13.209 ns |  13.383 ns |  1.00 |    0.00 |      - |         - |          NA |
|                                        Property_Set_class | Job-UYFZWZ | \INVOKE_BEFORE\corerun.exe |  18.799 ns | 0.1101 ns | 0.1030 ns |  18.764 ns |  18.636 ns |  19.001 ns |  1.41 |    0.01 |      - |         - |          NA |

Benchmarks for NativeAot. ~1.3-1.7x faster when using the new APIs

old 0 args: 1014
new 0 args:  595
old 1 arg: 4860
new 1 arg: 3319
old 2 args:7595
new 2 args:5875
old 5 args:16292
new 5 args:15325

[dotnet-issue-labeler]

Note regarding the new-api-needs-documentation label:

This serves as a reminder for when your PR is modifying a ref *.cs file and adding/modifying public APIs, please make sure the API implementation in the src *.cs file is documented with triple slash comments, so the PR reviewers can sign off that change.

[ghost]

Tagging subscribers to this area: @dotnet/area-system-reflection
See info in area-owners.md if you want to be subscribed.

Issue Details

---

[wip]

Author:	steveharter
Assignees:	steveharter
Labels:	area-System.Reflection
Milestone:	-

[buyaa-n]

Left a few NITs, overall looks good to me.

[buyaa-n]

NIT: By assuming this applyes to all new Invoke methods, this might be good to be added to the public doc

[cincuranet]

Improvement in dotnet/perf-autofiling-issues#19968.