Fix flakiness in CallChainReentrancy_WithSuppression

[ReubenBond]

Summary

• fix a race in CallChainReentrancyGrain.CallChain by snapshotting _unblocker at invocation start
• use that snapshot for all Task.WhenAny waits so in-flight calls are not affected by _unblocker reset timing

Root cause

UnblockWaiters signals and then replaces _unblocker. Depending on timing, an in-flight CallChain could read the new unsignaled task, causing an intermittent hang and eventual cancellation while waiting for (Exit, b, 4).

Testing

• dotnet test test/TesterInternal/TesterInternal.csproj --filter FullyQualifiedName~UnitTests.General.AllowCallChainReentrancyTests.CallChainReentrancy_WithSuppression
• 50 repeated runs on net8.0
• 50 repeated runs on net10.0
• AllowCallChainReentrancyTests on both net8.0 and net10.0

Microsoft Reviewers: Open in CodeFlow

[Copilot]

Pull request overview

This PR fixes a race condition in CallChainReentrancyGrain.CallChain that caused intermittent test failures in CallChainReentrancy_WithSuppression. The race occurred when UnblockWaiters replaced _unblocker while CallChain invocations were still in progress. The fix snapshots _unblocker at the start of each CallChain invocation to ensure consistent behavior throughout the call.

Changes:

• Added local snapshot of _unblocker at the start of CallChain method
• Updated all three Task.WhenAny calls to use the snapshotted value instead of the field