Move some folding for add/sub to gtFoldExpr

[tannergooding]

Continuation of #128187, but this time covering some simplifications for add/sub operations.

[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

Continuation of #128187 — moves several morph-time simplifications for ADD/SUB/NEG/NOT (double-negation removal, (-a)+b => b-a, a+(-b) => a-b, a-(-b) => a+b, (-a)-(-b) => b-a) out of fgMorphSmpOp/fgOptimizeAddition into a refactored gtFoldExpr. The dispatch in gtFoldExpr is split into new gtFoldExprUnary / gtFoldExprBinary helpers, and a new GenTree::ToggleReverseOp helper is introduced so swaps can preserve evaluation order without consulting gtCanSwapOrder.

Changes:

• Refactor gtFoldExpr into per-shape (Unary/Binary/Special) dispatch and add the corresponding folds for GT_NEG/GT_NOT and GT_ADD/GT_SUB with NEG operands.
• Remove the corresponding folds from fgMorphSmpOp and fgOptimizeAddition.
• Add GenTree::ToggleReverseOp() to flip GTF_REVERSE_OPS after swapping children.

Reviewed changes

Copilot reviewed 4 out of 4 changed files in this pull request and generated 3 comments.

File	Description
src/coreclr/jit/gentree.cpp	Refactors gtFoldExpr and adds new gtFoldExprUnary/gtFoldExprBinary with NEG/NOT/ADD/SUB folds; contains compile errors due to misspelled local references (op1Op1/op2Op1 vs op1op1/op2op1).
src/coreclr/jit/gentree.h	Adds ToggleReverseOp() helper to flip GTF_REVERSE_OPS.
src/coreclr/jit/compiler.h	Declares the new gtFoldExprUnary/gtFoldExprBinary helpers.
src/coreclr/jit/morph.cpp	Removes the simplifications now handled by gtFoldExprBinary/gtFoldExprUnary.

[Copilot]

Pull request overview

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

[Copilot]

Pull request overview

Copilot reviewed 3 out of 3 changed files in this pull request and generated 1 comment.

[Copilot]

Pull request overview

Copilot reviewed 3 out of 3 changed files in this pull request and generated 4 comments.

[Copilot]

Pull request overview

Copilot reviewed 4 out of 4 changed files in this pull request and generated 1 comment.

[Copilot]

Pull request overview

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

[Copilot]

Pull request overview

Copilot reviewed 4 out of 4 changed files in this pull request and generated 2 comments.

Comments suppressed due to low confidence (1)

src/coreclr/jit/morph.cpp:7867

• Removing fgPushConstantsRight from these compare cases means the subsequent const-specific morph optimizations now only run when the constant is already in op2. For patterns that commonly arrive as CONST relop x (e.g., source-level 0 == x), this can skip fgOptimizeEqualityComparisonWithConst/fgOptimizeRelationalComparisonWithConst in Tier0/minopts (gtFoldExprSpecial won’t run there to normalize). Consider pushing constants right here again (or teaching the const optimizers to handle a constant in either operand) so these optimizations don’t depend on prior canonicalization.

        case GT_EQ:
        case GT_NE:
        {
            oper = tree->OperGet();
            op1  = tree->gtGetOp1();
            op2  = tree->gtGetOp2();

            if (op2->IsIntegralConst())
            {
                tree = fgOptimizeEqualityComparisonWithConst(tree->AsOp());