MCP clients can timeout / cancel initiated requests. Currently in our tools, even after a timeout we don't cancel in-flight request to MongoDB server. So essentially even though client has timed out the request, server still continue to process the ongoing query.

We should ensure that all the in-flight requests to Atlas API / MongoDB server, adheres to MCP client request timeouts.

This can be facilitated using AbortSignal available on tool's execute method. We should pass down this abort signal to all the interfaces that generates IO requests.