fix(ingest): avoid multiprocessing "fork" start method

[hsheth2]

We use threads in a variety of places within our CLI. The multiprocessing "fork" start method is not safe to use with threads. MacOS and Windows already default to "spawn", and Linux will as well starting in Python 3.14. See https://docs.python.org/3/library/multiprocessing.html#contexts-and-start-methods

This fixes a bug where the classification module would deadlock when using more than one worker process. Reverts #12261, since that workaround is no longer required.

By making changes to both the entrypoint and the classification module, we can ensure that

1. if people use our code via Pipeline.create(...).run(), we will not cause any deadlocks
2. we will play nice with any dependencies that use multiprocessing

Eventually it may make sense to use "forkserver" as the default where available, but we can revisit that in the future.

Tested locally.

Checklist

• The PR conforms to DataHub's Contributing Guideline (particularly Commit Message Format)
• Links to related issues (if applicable)
• Tests for the changes have been added/updated (if applicable)
• Docs related to the changes have been added/updated (if applicable). If a new feature has been added a Usage Guide has been added for the same.
• For any breaking change/potential downtime/deprecation/big changes an entry has been made in Updating DataHub

[codecov]

Codecov Report

Attention: Patch coverage is 75.00000% with 1 line in your changes missing coverage. Please review.

✅ All tests successful. No failed tests found.

Files with missing lines	Patch %	Lines
metadata-ingestion/src/datahub/entrypoints.py	50.00%	1 Missing ⚠️

❌ Your patch status has failed because the patch coverage (75.00%) is below the target coverage (90.00%). You can increase the patch coverage or adjust the target coverage.

Files with missing lines	Coverage Δ
...datahub/ingestion/glossary/classification_mixin.py	89.14% <100.00%> (+5.55%)	⬆️
...stion/src/datahub/ingestion/glossary/classifier.py	95.65% <100.00%> (+0.19%)	⬆️
metadata-ingestion/src/datahub/entrypoints.py	68.57% <50.00%> (-0.27%)	⬇️

... and 14 files with indirect coverage changes

---

Continue to review full report in Codecov by Sentry.

Legend - Click here to learn more
Δ = absolute <relative> (impact), ø = not affected, ? = missing data
Powered by Codecov. Last update 6f0d475...142cac3. Read the comment docs.