Implement file-level parallelization

[drdavella]

Overview

Implement (optional) per-codemod file-level parallelization

Description

• This PR enables files to be processed in parallel for each codemod
• There is no parallelism between codemods since we need to preserve strict ordering
• The default behavior for now is to use a single thread, so there is no difference from the previous behavior
• All per-file codemod state is now managed by FileContext and then aggregated at the end of each codemod pass
• This will enable us to perform testing in more resource constrained environments where more threads might be useful. In particular, we suspect that environments that are highly limited by IO bandwidth will benefit from parallelization

[codecov]

Codecov Report

Merging #100 (15716b7) into main (73f5b4a) will decrease coverage by 0.10%.
The diff coverage is 100.00%.

❗ Current head 15716b7 differs from pull request most recent head 43b88ab. Consider uploading reports for the commit 43b88ab to get more accurate results

@@            Coverage Diff             @@
##             main     #100      +/-   ##
==========================================
- Coverage   95.63%   95.54%   -0.10%     
==========================================
  Files          60       60              
  Lines        2451     2446       -5     
==========================================
- Hits         2344     2337       -7     
- Misses        107      109       +2     

Files	Coverage Δ
src/codemodder/cli.py	100.00% <100.00%> (ø)
src/codemodder/codemodder.py	98.14% <100.00%> (+0.08%)	⬆️
src/codemodder/codemods/api/__init__.py	95.31% <100.00%> (-0.08%)	⬇️
src/codemodder/codemods/base_codemod.py	100.00% <100.00%> (ø)
src/codemodder/context.py	97.53% <100.00%> (+0.23%)	⬆️
src/codemodder/file_context.py	100.00% <100.00%> (ø)
...codemodder/project_analysis/python_repo_manager.py	100.00% <100.00%> (ø)
src/core_codemods/order_imports.py	92.85% <ø> (ø)
src/core_codemods/sql_parameterization.py	91.32% <100.00%> (-0.04%)	⬇️
src/core_codemods/upgrade_sslcontext_tls.py	100.00% <100.00%> (ø)

... and 1 file with indirect coverage changes

[clavedeluna]

How much of a PITA would it be to add an integration test with --max-workers used?

[drdavella]

@clavedeluna since the default is to use one worker and since this is purely to enable performance testing I'm going to skip the integration test for now. Once we decide we want to use --max-workers > 1 in production environments we can revisit with an integration test.