Add assembly version of simple operations on aarch64

For aarch64 and arm64ec with Neon, add assembly versions of the
following:

* `ceil`
* `ceilf`
* `fabs`
* `fabsf`
* `floor`
* `floorf`
* `fma`
* `fmaf`
* `round`
* `roundf`
* `sqrt`
* `sqrtf`
* `trunc`
* `truncf`

If the `fp16` target feature is available, which implies `neon`, also
include the following:

* `ceilf16`
* `fabsf16`
* `floorf16`
* `rintf16`
* `roundf16`
* `sqrtf16`
* `truncf16`

Additionally, replace `core::arch` versions of the following with
handwritten assembly (which avoids issues with `aarch64be`):

* `rint`
* `rintf`

Instructions for `fmax` and `fmin` are also available but seem to
provide different results based on whether NaN inputs are signaling or
quiet. Our current implementation does not do this, so omit these for
now.

Author: tgross35

etc/function-definitions.json

@@ -107,6 +107,7 @@
     "ceil": {
         "sources": [
             "src/libm_helper.rs",
+            "src/math/arch/aarch64.rs",
             "src/math/arch/i586.rs",
             "src/math/arch/wasm32.rs",
             "src/math/ceil.rs",
@@ -116,6 +117,7 @@
     },
     "ceilf": {
         "sources": [
+            "src/math/arch/aarch64.rs",
             "src/math/arch/wasm32.rs",
             "src/math/ceilf.rs",
             "src/math/generic/ceil.rs"
@@ -131,6 +133,7 @@
     },
     "ceilf16": {
         "sources": [
+            "src/math/arch/aarch64.rs",
             "src/math/ceilf16.rs",
             "src/math/generic/ceil.rs"
         ],
@@ -274,6 +277,7 @@
     "fabs": {
         "sources": [
             "src/libm_helper.rs",
+            "src/math/arch/aarch64.rs",
             "src/math/arch/wasm32.rs",
             "src/math/fabs.rs",
             "src/math/generic/fabs.rs"
@@ -282,6 +286,7 @@
     },
     "fabsf": {
         "sources": [
+            "src/math/arch/aarch64.rs",
             "src/math/arch/wasm32.rs",
             "src/math/fabsf.rs",
             "src/math/generic/fabs.rs"
@@ -297,6 +302,7 @@
     },
     "fabsf16": {
         "sources": [
+            "src/math/arch/aarch64.rs",
             "src/math/fabsf16.rs",
             "src/math/generic/fabs.rs"
         ],
@@ -334,6 +340,7 @@
     "floor": {
         "sources": [
             "src/libm_helper.rs",
+            "src/math/arch/aarch64.rs",
             "src/math/arch/i586.rs",
             "src/math/arch/wasm32.rs",
             "src/math/floor.rs",
@@ -343,6 +350,7 @@
     },
     "floorf": {
         "sources": [
+            "src/math/arch/aarch64.rs",
             "src/math/arch/wasm32.rs",
             "src/math/floorf.rs",
             "src/math/generic/floor.rs"
@@ -358,6 +366,7 @@
     },
     "floorf16": {
         "sources": [
+            "src/math/arch/aarch64.rs",
             "src/math/floorf16.rs",
             "src/math/generic/floor.rs"
         ],
@@ -366,12 +375,14 @@
     "fma": {
         "sources": [
             "src/libm_helper.rs",
+            "src/math/arch/aarch64.rs",
             "src/math/fma.rs"
         ],
         "type": "f64"
     },
     "fmaf": {
         "sources": [
+            "src/math/arch/aarch64.rs",
             "src/math/fmaf.rs"
         ],
         "type": "f32"
@@ -725,6 +736,7 @@
     },
     "rintf16": {
         "sources": [
+            "src/math/arch/aarch64.rs",
             "src/math/generic/rint.rs",
             "src/math/rintf16.rs"
         ],
@@ -733,13 +745,15 @@
     "round": {
         "sources": [
             "src/libm_helper.rs",
+            "src/math/arch/aarch64.rs",
             "src/math/generic/round.rs",
             "src/math/round.rs"
         ],
         "type": "f64"
     },
     "roundf": {
         "sources": [
+            "src/math/arch/aarch64.rs",
             "src/math/generic/round.rs",
             "src/math/roundf.rs"
         ],
@@ -754,6 +768,7 @@
     },
     "roundf16": {
         "sources": [
+            "src/math/arch/aarch64.rs",
             "src/math/generic/round.rs",
             "src/math/roundf16.rs"
         ],
@@ -816,6 +831,7 @@
     "sqrt": {
         "sources": [
             "src/libm_helper.rs",
+            "src/math/arch/aarch64.rs",
             "src/math/arch/i686.rs",
             "src/math/arch/wasm32.rs",
             "src/math/generic/sqrt.rs",
@@ -825,6 +841,7 @@
     },
     "sqrtf": {
         "sources": [
+            "src/math/arch/aarch64.rs",
             "src/math/arch/i686.rs",
             "src/math/arch/wasm32.rs",
             "src/math/generic/sqrt.rs",
@@ -841,6 +858,7 @@
     },
     "sqrtf16": {
         "sources": [
+            "src/math/arch/aarch64.rs",
             "src/math/generic/sqrt.rs",
             "src/math/sqrtf16.rs"
         ],
@@ -888,6 +906,7 @@
     "trunc": {
         "sources": [
             "src/libm_helper.rs",
+            "src/math/arch/aarch64.rs",
             "src/math/arch/wasm32.rs",
             "src/math/generic/trunc.rs",
             "src/math/trunc.rs"
@@ -896,6 +915,7 @@
     },
     "truncf": {
         "sources": [
+            "src/math/arch/aarch64.rs",
             "src/math/arch/wasm32.rs",
             "src/math/generic/trunc.rs",
             "src/math/truncf.rs"
@@ -911,6 +931,7 @@
     },
     "truncf16": {
         "sources": [
+            "src/math/arch/aarch64.rs",
             "src/math/generic/trunc.rs",
             "src/math/truncf16.rs"
         ],