Fix stackoverflowtester to skip ARM32 in TestStackOverflowLargeFrameMainThread (#126525)

main PR <!-- Link to PR if any that fixed this in the main branch. -->

# Description

`TestStackOverflowLargeFrameMainThread` was missing `Architecture.Arm`
in its Unix skip guard, causing it to run on ARM32 where it reliably
fails. On ARM32, large-frame methods (65KB/frame) cause the JIT to probe
stack pages without moving SP, so the guard page fault lands >8KB from
SP — outside the runtime's 2-page detection window in `signal.cpp`. This
produces SIGSEGV (exit `0x8B`/139) instead of the expected SIGABRT (exit
`0x86`/134).

`TestStackOverflowLargeFrameSecondaryThread` already correctly included
`Architecture.Arm` in its skip. This change makes the main-thread
version consistent:

```csharp
// Before
if (((RuntimeInformation.ProcessArchitecture == Architecture.Arm64) || ... ||
    (RuntimeInformation.ProcessArchitecture == Architecture.LoongArch64)) && ...)

// After
if (((RuntimeInformation.ProcessArchitecture == Architecture.Arm64) || ... ||
    (RuntimeInformation.ProcessArchitecture == Architecture.LoongArch64) ||
    (RuntimeInformation.ProcessArchitecture == Architecture.Arm)) && ...)
```

# Customer Impact

No customer impact — this is a test-only fix. The underlying ARM32 stack
overflow detection limitation (tracked in #110173) remains; this just
prevents a spurious CI failure.

# Regression

Not a regression in product behavior. The test skip was simply omitted
from `TestStackOverflowLargeFrameMainThread` while it was correctly
applied to the secondary-thread variant.

# Testing

The fix aligns the skip condition with the already-correct
`TestStackOverflowLargeFrameSecondaryThread`. No behavior change on
non-ARM32 platforms.

# Risk

Minimal. Test-only change that adds an early-return on ARM32 Unix,
matching existing patterns in the same file.

# Package authoring no longer needed in .NET 9

IMPORTANT: Starting with .NET 9, you no longer need to edit a NuGet
package's csproj to enable building and bump the version.
Keep in mind that we still need package authoring in .NET 8 and older
versions.

---------

Co-authored-by: copilot-swe-agent[bot] <198982749+Copilot@users.noreply.github.com>
Co-authored-by: mangod9 <61718172+mangod9@users.noreply.github.com>

Author: Copilot

src/tests/baseservices/exceptions/stackoverflow/stackoverflowtester.cs

@@ -124,14 +124,15 @@ public static void TestStackOverflowSmallFrameMainThread()
         public static void TestStackOverflowLargeFrameMainThread()
         {
             if (((RuntimeInformation.ProcessArchitecture == Architecture.Arm64) || (RuntimeInformation.ProcessArchitecture == Architecture.X64) || (RuntimeInformation.ProcessArchitecture == Architecture.RiscV64) ||
-                (RuntimeInformation.ProcessArchitecture == Architecture.LoongArch64)) &&
+                (RuntimeInformation.ProcessArchitecture == Architecture.LoongArch64) || (RuntimeInformation.ProcessArchitecture == Architecture.Arm)) &&
                 ((Environment.OSVersion.Platform == PlatformID.Unix) || (Environment.OSVersion.Platform == PlatformID.MacOSX)))
             {
-                // Disabled on Unix ARM64, X64, RISCV64, and LoongArch64.
+                // Disabled on Unix RISCV64 and LoongArch64, similar to ARM64.
                 // LoongArch64 hit this issue on Alpine. TODO: implement stack probing using helpers.
-                // ARM64 and X64 disabled due to https://github.com/dotnet/runtime/issues/13519
+                // Disabled on Unix ARM64 due to https://github.com/dotnet/runtime/issues/13519
                 // The current stack probing doesn't move the stack pointer and so the runtime sometimes cannot
                 // recognize the underlying sigsegv as stack overflow when it probes too far from SP.
+                // Disabled on Unix X64/Arm due to https://github.com/dotnet/runtime/issues/110173 which needs investigation.
                 return;
             }