Ensuring stdio transport closure despite server hanging. Fixing #271

[TomasRup]

Force closing the child process upon closing the stdio process

Motivation and Context

The issue was raised and reproducible in #271

How Has This Been Tested?

Added a manual test + tried adding the scenario from the issue with Microsoft's playwright-mcp

Breaking Changes

None

Types of changes

• Bug fix (non-breaking change which fixes an issue)
• New feature (non-breaking change which adds functionality)
• Breaking change (fix or feature that would cause existing functionality to change)
• Documentation update

Checklist

• I have read the MCP Documentation
• My code follows the repository's style guidelines
• New and existing tests pass locally
• I have added appropriate error handling
• I have added or updated documentation as needed

Additional context

• StdioClientTransport#close() does not wait until underlying process is closed #271
• MCP server hangs when MCP client calls close() before browser_close tool is called microsoft/playwright-mcp#141

[Skn0tt]

Sending SIGKILL immediately doesn't allow the server to perform graceful shutdown, I don't think we should do that. So let's give the process a couple of seconds to exit by itself, and only send SIGKILL if it hangs.

[TomasRup]

@Skn0tt thanks, you are right, pushed a more graceful exit.

[Skn0tt]

After sleeping a couple nights over this, i'm not convinced this PR is the right approach. The gist of #271 is that close() doesn't block until the process is closed. This PR doesn't adress that, but instead adds graceful shutdown, which is a different issue. Before we implement graceful shutdown, I think we should coordinate with the project maintainers in a separate issue.