Ensure we don't do duplicate checks for various multi-reg paths

[tannergooding]

This had a significant improvement to TP for Arm64 in the other PR where I had attempted to ensure other reg state was consistently cloned.

[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

This PR refactors multi-reg handling in the JIT GenTree helpers to reduce redundant multi-reg classification checks, and adjusts debug tree-dump helpers to work independent of FEATURE_MULTIREG_RET.

Changes:

• Refactor GenTree::GetRegisterDstCount to directly handle known multi-reg shapes (calls, copy/reload, HW intrinsics, multi-reg locals) without a leading IsMultiRegNode() gate.
• Minor cleanup in GenTree::IsMultiRegNode() to use IsCopyOrReload() instead of opcode checks.
• Remove FEATURE_MULTIREG_RET guards around debug display helpers (gtDispMultiRegCount, multi-reg printing in gtDispRegVal) and the corresponding declaration in compiler.h.

Reviewed changes

Copilot reviewed 2 out of 2 changed files in this pull request and generated 2 comments.

File	Description
src/coreclr/jit/gentree.cpp	Refactors multi-reg destination count logic and adjusts debug multi-reg display helpers / guards.
src/coreclr/jit/compiler.h	Makes gtDispMultiRegCount always declared in DEBUG builds (no FEATURE_MULTIREG_RET guard).

[tannergooding]

CC. @dotnet/jit-contrib, @jakobbotsch, @EgorBo. Small cleanup to various key multi-reg paths ensuring we aren't doing an expensive general purpose check and then repeating all the same checks for each path we need to handle.

This provides a good TP improvement for Arm64, giving us -0.32% to 0.09% in MinOpts and -0.08% to -0.02% in FullOpts, with the largest improvements being in the realworld collection.