Remove bounds check for arr1[arr2.Length] pattern

[EgorBo]

Closes #115793

static bool Repro(string prefix, string path)
{
    if (prefix.Length < path.Length)
        return (path[prefix.Length] == '/');

    return false;
}

 ; Method Program:Repro(System.String,System.String):ubyte (FullOpts)
-      sub      rsp, 40
       mov      ecx, dword ptr [rcx+0x08]
-      mov      r8d, dword ptr [rdx+0x08]
-      cmp      ecx, r8d
+      mov      eax, dword ptr [rdx+0x08]
+      cmp      ecx, eax
       jl       SHORT G_M15110_IG05
       xor      eax, eax
-      add      rsp, 40
       ret      
 G_M15110_IG05:
-      cmp      ecx, r8d
-      jae      SHORT G_M15110_IG07
       mov      eax, ecx
       cmp      word  ptr [rdx+2*rax+0x0C], 47
       sete     al
       movzx    rax, al
-      add      rsp, 40
       ret      
-G_M15110_IG07:
-      call     CORINFO_HELP_RNGCHKFAIL
-      int3     
-; Total bytes of code: 53
+; Total bytes of code: 28

A few diffs

[Copilot]

Pull Request Overview

This PR updates the JIT range-check logic to recognize the new GT_ARR_LENGTH node and remove an unnecessary bounds check for patterns like arr1[arr2.Length], improving codegen size and performance.

• Treat GT_ARR_LENGTH as non-overflowing in ComputeDoesOverflow
• Provide a concrete range (0 to CORINFO_Array_MaxLength) for GT_ARR_LENGTH in ComputeRange
• Eliminate redundant range‐check helper paths, reducing code size

Comments suppressed due to low confidence (1)

src/coreclr/jit/rangecheck.cpp:1647

• Consider adding unit tests for the GT_ARR_LENGTH case to validate that the computed range (0 to CORINFO_Array_MaxLength) behaves as expected and guard against future regressions in this scenario.

else if (expr->OperIs(GT_ARR_LENGTH))

[EgorBo]

PTAL @jakobbotsch @dotnet/jit-contrib small change, a couple of diffs

[EgorBo]

/ba-g "NativeAOT timeouts"