Fix SIGINT handler for StopOnExitJobServer

[shunping]

The previously registered SIGINT handler had an incorrect signature according to the signal module documentation:
https://docs.python.org/3/library/signal.html#signal.signal

The handler is called with two arguments: the signal number and the current stack frame (None or a frame object; for a description of frame objects,

This PR updates the handler to the correct signature. It now gracefully stops the server and ensures the original handler is called for proper application exit. This is important especially in a Colab environment, because the kernel will send out SIGINT when the stop button of a cell is pressed.

[lostluck]

That definitely explains the exception message. Good find!

[shunping]

Run Python_Dataframes PreCommit 3.11

[codecov]

Codecov Report

Attention: Patch coverage is 50.00000% with 3 lines in your changes missing coverage. Please review.

Project coverage is 56.50%. Comparing base (8cd0a29) to head (530fab6).
Report is 12 commits behind head on master.

Files with missing lines	Patch %	Lines
...thon/apache_beam/runners/portability/job_server.py	50.00%	3 Missing ⚠️

Additional details and impacted files

@@             Coverage Diff              @@
##             master   #34748      +/-   ##
============================================
+ Coverage     56.48%   56.50%   +0.02%     
  Complexity     3289     3289              
============================================
  Files          1178     1180       +2     
  Lines        180978   181262     +284     
  Branches       3399     3399              
============================================
+ Hits         102220   102419     +199     
- Misses        75499    75584      +85     
  Partials       3259     3259              

Flag	Coverage Δ
python	81.26% <50.00%> (-0.05%)	⬇️

Flags with carried forward coverage won't be shown. Click here to find out more.

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.
• 📦 JS Bundle Analysis: Save yourself from yourself by tracking and limiting bundle sizes in JS merges.

[lostluck]

Does this help with #33623 in combination with the Singleton experiment?

[github-actions]

Assigning reviewers. If you would like to opt out of this review, comment assign to next reviewer:

R: @claudevdm for label python.

Available commands:

• stop reviewer notifications - opt out of the automated review tooling
• remind me after tests pass - tag the comment author after tests pass
• waiting on author - shift the attention set back to the author (any comment or push by the author will return the attention set to the reviewers)

The PR bot will only process comments in the main thread (not review comments).

[shunping]

Run Python_Runners PreCommit 3.9

[shunping]

Does this help with #33623 in combination with the Singleton experiment?

Yes, there are multiple changes recently to address #33623.

• Use cached prism binary without re-unzipping by default. #34616: fixed the cached mechanism so that we won't try to overwrite a running prism binary if the binary is in cache
• Add a md5sum check before overwriting a cached prism file #34715: supported a remote filesystem path in --prism_location. Note that in this case by design, cache will be disabled, so a proper md5sum checking is implemented so that we won't overwrite a running prism binary if the remote file has the same md5 as the cached one.
• Add an experiment to make prism server a singleton #34623: In the case of more than one pipelines running before the interpreter (or the kernel in colab) terminates, we will only start one prism server and run all pipelines with it. This is useful for running batch pipelines in Colab where pipelines (and the cells they are in) will run and finish by themselves.
• Fix SIGINT handler for StopOnExitJobServer #34748: This PR. It fixes a bug in SIGINT handler, which is invoked when we run a streaming pipeline and click the stop button in Colab.

'PrismJobServer' is wrapped around StopOnExitJobServer', which is then made into a singleton. After this PR, when the event of SIGINT (eg. pressing the stop button in Colab or Ctrl + c in terminal) is triggered, the prism server will stop gracefully rather than being defunct'ed.

The singleton of 'PrismJobServer' on the other hand is still valid (though there is no prism subprocess running atm), and we will be able to start the subprocess again if we run another pipeline on the same Colab.

The current behavior is the same for both Colab and Terminal. A future improvement is that we can keep the prism server alive on the event of SIGINT in an interactive environment and, as @lostluck suggested in another thread, enforce an idle timeout.

[lostluck]

Fantastic!