fix: catch ClosedResourceError when sending error log to disconnected client

[giulio-leone]

Summary

Fixes #2064

When a client disconnects during request handling in a stateless streamable-HTTP server, _handle_message's error handler tries to call send_log_message() back to the client. Since the write stream is already closed, this raises ClosedResourceError, which is unhandled and crashes the stateless session with an ExceptionGroup.

Root Cause

This is a different code path from what PR #1384 fixed. That PR addressed ClosedResourceError in the message router loop. This bug is in the error recovery path:

ClientDisconnect → _handle_message catches Exception
  → tries send_log_message() to notify client
  → write stream already closed → ClosedResourceError
  → unhandled in TaskGroup → ExceptionGroup crash

Fix

Added try/except around send_log_message() in _handle_message to catch both ClosedResourceError and BrokenResourceError, logging a debug message instead of crashing. Failing to notify a disconnected client is expected and harmless.

Tests

Added 3 tests to test_lowlevel_exception_handling.py:

• test_exception_handling_tolerates_closed_write_stream[closed] — ClosedResourceError
• test_exception_handling_tolerates_closed_write_stream[broken] — BrokenResourceError
• test_exception_handling_reraises_with_closed_stream — original exception still re-raised when raise_exceptions=True

All 9 tests pass.

[giulio-leone]

Friendly ping — CI is green and this is ready for review. Happy to address any feedback. Thanks!

[giulio-leone]

All CI checks pass. Ready for review.