Expose internal stack introspection APIs as a statistical runtime analysis tool

[pablogsal]

Feature or enhancement

Implement a new high-frequency runtime observer that leverages the existing _remote_debugging module to collect detailed execution statistics from running Python processes. This tool enables non-intrusive observation of production applications by attaching to already-running processes without requiring any code modifications, restarts, or special startup flags. The observer can perform extremely high-frequency sampling of stack traces and interpreter state, providing detailed runtime execution analysis of live applications.

Linked PRs

• gh-135953: Implement sampling tool under profile.sample #135998
• gh-135953: Profile a module or script with sampling profiler #136777
• gh-135953: Add flamegraph reporter to sampling profiler #138304
• gh-135953: Add flamegraph reporter to sampling profiler #138715
• gh-135953: Fix refleak in cache method #138721
• gh-135953: Reduce memory usage of stack collectors #138875
• gh-135953: Prevent overflow in the profiler flamegraph hover boxes #138880
• gh-135953: Add Gecko reporter to sampling profiler #139364
• gh-135953: Add GIL contention markers to sampling profiler Gecko format #139485

[pablogsal]

I'm working on this with @lkollar