[arm64] Addressing modes for SIMD

[EgorBo]

This PR implements addressing modes for SIMD vectors on arm64 (for now only loads with unscaled indices)
Closes #67435

Benchmark:

static byte[] _array1 = new byte[100000]; 
static byte[] _array2 = new byte[100000]; 
 
[Benchmark] 
public bool SequenceEqual() => _array1.AsSpan().SequenceEqual(_array2);

Method	Toolchain	Mean	Error	StdDev	Ratio
SequenceEqual	/Core_Root/corerun	3.635 us	0.0031 us	0.0027 us	1.00
SequenceEqual	/Core_Root_pr/corerun	2.546 us	0.0224 us	0.0210 us	0.70

codegen diff example: https://www.diffchecker.com/wACQMKBT (SpanHelpers.SequenceEqual)

[ghost]

Tagging subscribers to this area: @JulieLeeMSFT
See info in area-owners.md if you want to be subscribed.

Issue Details

---

This PR implements addressing modes for SIMD vectors on arm64

Benchmark:

static byte[] _array1 = new byte[100000]; 
static byte[] _array2 = new byte[100000]; 
 
[Benchmark] 
public bool SequenceEquals() => _array1.AsSpan().SequenceEqual(_array2);

Method	Toolchain	Mean	Error	StdDev	Ratio
SequenceEquals	/Core_Root/corerun	3.635 us	0.0031 us	0.0027 us	1.00
SequenceEquals	/Core_Root_pr/corerun	2.546 us	0.0224 us	0.0210 us	0.70

codegen diff example: https://www.diffchecker.com/wACQMKBT (SpanHelpers.SequenceEqual)

Author:	EgorBo
Assignees:	EgorBo
Labels:	area-CodeGen-coreclr
Milestone:	-

[EgorBo]

@dotnet/jit-contrib PTAL, simple change, diffs are small but cover SpanHelpers methods which are perf-critical

[danmoseley]

Is this change sufficiently protected by an existing benchmark?

[EgorBo]

Is this change sufficiently protected by an existing benchmark?

Yes, SpanHelpers are heavily used under the hood in many benchmarks and Regexes, we even have explicit benchmarks, e.g. https://github.com/dotnet/performance/blob/d7dac8a7ca12a28d099192f8a901cf8e30361384/src/benchmarks/micro/libraries/System.Memory/Span.cs#L48