JIT: Remove dead Vector2/3 lowering code

[saucecontrol]

Since the old SIMD types now use HWIntrinsics, these paths are dead.

[dotnet-policy-service]

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

[Copilot]

Pull request overview

Removes lowering/codegen special-cases for 8/12-byte SIMD shapes (Vector2/Vector3) in JIT HWIntrinsic paths, based on the premise that these shapes are no longer produced now that the relevant SIMD types are implemented via HWIntrinsics.

Changes:

• Simplify LowerHWIntrinsicCmpOp and LowerHWIntrinsicDot by removing SIMD8/SIMD12-specific logic and tightening associated assertions.
• Tighten constant-create assertions to exclude SIMD8/SIMD12 where those sizes are no longer expected.
• Use ToScalar instead of GetElement(0) for ARM64 Dot result extraction.

Reviewed changes

Copilot reviewed 4 out of 4 changed files in this pull request and generated no comments.

File	Description
src/coreclr/jit/lowerxarch.cpp	Removes SIMD8/SIMD12 handling in compare/dot lowering and tightens create assertions.
src/coreclr/jit/lowerarmarch.cpp	Removes SIMD12-specific compare/dot handling and tightens create assertions; updates comment to Vector64/128.
src/coreclr/jit/hwintrinsicarm64.cpp	Changes Dot scalar extraction to use gtNewSimdToScalarNode.
src/coreclr/jit/gentree.cpp	Tightens comparison intrinsic size assertion (drops SIMD12).

[saucecontrol]

cc @dotnet/jit-contrib

This is zero diff for aarch, small improvement for xarch due to removal of the containment check bailout for SIMD8 and SIMD12 (these are handled correctly now that all HWIntrinsic containment logic uses operand size).

Diffs

[Copilot]

Pull request overview

Copilot reviewed 4 out of 4 changed files in this pull request and generated no new comments.

[tannergooding]

CC. @kg, @EgorBo for secondary review