remove global multiprocessing.set_start_method

[braingram]

Fixes: #8306

Without setting a multiprocessing start method (using the default) on linux with python 3.12 results in a DeprecationWarning.

Warnings for python 3.12:

  jwst/jump/tests/test_detect_jumps.py: 9 warnings
  jwst/jump/tests/test_jump_step.py: 24 warnings
    /opt/hostedtoolcache/Python/3.12.2/x64/lib/python3.12/multiprocessing/popen_fork.py:66: DeprecationWarning: This process (pid=2483) is multi-threaded, use of fork() may lead to deadlocks in the child.
      self.pid = os.fork()
  
  jwst/ramp_fitting/tests/test_ramp_fit.py::test_multiprocessing
  jwst/ramp_fitting/tests/test_ramp_fit.py::test_multiprocessing2
  jwst/ramp_fitting/tests/test_ramp_fit_step.py::test_ramp_fit_step[2]
  jwst/ramp_fitting/tests/test_ramp_fit_step.py::test_ramp_fit_step[2]
  jwst/ramp_fitting/tests/test_ramp_fit_step.py::test_ramp_fit_step[half]
  jwst/ramp_fitting/tests/test_ramp_fit_step.py::test_ramp_fit_step[half]
  jwst/ramp_fitting/tests/test_ramp_fit_step.py::test_ramp_fit_step[all]
  jwst/ramp_fitting/tests/test_ramp_fit_step.py::test_ramp_fit_step[all]
    /opt/hostedtoolcache/Python/3.12.2/x64/lib/python3.12/multiprocessing/popen_fork.py:66: DeprecationWarning: This process (pid=2486) is multi-threaded, use of fork() may lead to deadlocks in the child.
      self.pid = os.fork()

This is due to fork being unsafe for applications that use threads. This warning was previously silenced using a global set_start_method which can cause issues for other packages (see #8306).

stcal (which is the primary user of multiprocessing in the pipeline) was updated:
spacetelescope/stcal#249

This PR temporarily uses stcal main (that can be removed prior to merge or after a stcal release is made) to pick up the changes in the above PR and updates the remaining multiprocessing usage in wfss_contam to only set forkserver in a context (to not impact other packages). EDIT: stcal 1.7.0 (now the required version) contains the required changes

[codecov]

Codecov Report

Attention: Patch coverage is 33.33333% with 2 lines in your changes are missing coverage. Please review.

Project coverage is 75.77%. Comparing base (2fb073e) to head (18344af).
Report is 27 commits behind head on master.

Files	Patch %	Lines
jwst/wfss_contam/observations.py	33.33%	2 Missing ⚠️

Additional details and impacted files

@@            Coverage Diff             @@
##           master    #8343      +/-   ##
==========================================
+ Coverage   75.31%   75.77%   +0.46%     
==========================================
  Files         474      476       +2     
  Lines       38965    39448     +483     
==========================================
+ Hits        29345    29891     +546     
+ Misses       9620     9557      -63     

Flag	Coverage Δ		*Carryforward flag
nightly	77.66% <ø> (+0.32%)	⬆️	Carriedforward from 2fb073e

*This pull request uses carry forward flags. Click here to find out more.

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

[braingram]

Regtests run at: https://plwishmaster.stsci.edu:8081/job/RT/job/JWST-Developers-Pull-Requests/1310/
26 failures which appear to match the 26 on main:
https://plwishmaster.stsci.edu:8081/blue/organizations/jenkins/RT%2FJWST/detail/JWST/2808/tests
and match up with the description of the crds update today.

[braingram]

Note: prior to merge the dev dependency on stcal should be removed.

[braingram]

The 3.9 and oldestdeps failures look like they'd be addressed in:
#8365

[braingram]

@emolter this PR has some minor overlap with #8417 I gave #8417 a skim and it looks like it would make the wfss changes in this PR unecessary. However this PR has additional related changes in jump and ramp_fitting. What would you recommend for sequencing these? As the changes in this PR are minor compared to #8417 can we get this one merged first? I'll rebase and let me know if you'd like to see a newer regtest run.

[emolter]

The code changes look good to me. Are there any regtests that test the multiprocessing modes for these steps? Not sure what that would even look like...

RE my PR for the wfss_contam step, I think it's completely fine to merge this right away. It will be a while before that PR gets merged as there are lots of outstanding issues.

[braingram]

Thanks for taking a look!

The code changes look good to me. Are there any regtests that test the multiprocessing modes for these steps? Not sure what that would even look like...

Not that I'm aware of. The main motivation for adding the set_start_method in #8093 is the DeprecationWarnings that will be seen if using fork (the default on linux) and python 3.12. I'm happy to go into more details but the tldr is that we need to set a non-fork method to avoid these warnings.

RE my PR for the wfss_contam step, I think it's completely fine to merge this right away. It will be a while before that PR gets merged as there are lots of outstanding issues.

Good to know. Let me if there's anything I can do to help.

[hbushouse]

Looks good to me. Is everyone else happy?