Expose new JITModule::read_got_entry function

[eggyal]

This hasn't been discussed in a wasmtime issue, but (together with #2786) it's needed it for bjorn3/rustc_codegen_cranelift#1166.

It exposes JITModule::get_got_address a new JITModule::read_got_entry function so that, when jitting lazily, cranelift users can inspect the GOT to determine whether a given function's entry is still their shim trampoline or not (and therefore whether the function has already been jitted). This prevents the function from being jitted differently from simultaneous requests arising from different threads.

No test cases added, because exposing a function obviously doesn't change any behaviour.

Sorry, I don't know who would be an appropriate reviewer for this.

[bjorn3]

Instead of exposing get_got_address directly, can you expose a function accepting a FuncId that reads the got entry?

[bjorn3]

@cfallin Could you take a look at this?

[cfallin]

Some exciting but almost certainly unrelated CI failures on Windows -- corruption issue of some sort?

     Running target\release\deps\strings-bc2f113fd401e08a.exe

running 3 tests
test hello_string ... ok
test overlapping_string ... ok
test multi_string ... FAILED

failures:

---- multi_string stdout ----
proptest: FileFailurePersistence::SourceParallel set, but failed to find lib.rs or main.rs
len=839, a='Πῌ𝈍ᾲὴὝ𝈡ῳὼ΄𝈘ᾶΘΊἝΈͽͿΆΆᴨφͺὗΊͿ῁𐅔ὕΖϴὌῪ ...

(This seems to be in wiggle -- cc @pchickey @sunfishcode)

restarting CI.