chore(python): Revert jsonschema updates to fix snapshot tests

[SyntaxColoring]

Overview

Some analysis snapshot tests are failing indicating a jsonschema-related error. (e.g. #15750).

I think this is a problem introduced by commit 8f2583f, PR #15720. This PR reverts that until we understand more.

Test Plan

See: #15759 (comment)

Review requests

None.

Risk assessment

Who knows, at this point.

[codecov]

Codecov Report

All modified and coverable lines are covered by tests ✅

Project coverage is 63.68%. Comparing base (0a4b774) to head (f4766a7).
Report is 7 commits behind head on edge.

Additional details and impacted files

@@            Coverage Diff             @@
##             edge   #15759      +/-   ##
==========================================
- Coverage   63.70%   63.68%   -0.03%     
==========================================
  Files         300      300              
  Lines       15649    15653       +4     
==========================================
- Hits         9969     9968       -1     
- Misses       5680     5685       +5     

Flag	Coverage Δ
g-code-testing	92.43% <ø> (ø)
hardware	55.73% <100.00%> (-0.05%)	⬇️
shared-data	75.91% <ø> (+0.05%)	⬆️
system-server	94.02% <ø> (ø)

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

Files	Coverage Δ
...are/opentrons_hardware/drivers/can_bus/settings.py	95.77% <100.00%> (ø)

... and 10 files with indirect coverage changes

[sfoster1]

This looks good to me. That hardware testing failure is also in edge.

[SyntaxColoring]

Here's a command to help test this locally, which roughly mimics what analyses-snapshot-testing's Dockerfile does:

rm -rf venv &&
python -m venv venv &&
venv/bin/pip install -U ../shared-data/python '../hardware[flex]' ../api pandas==1.4.3 && 
venv/bin/python -m opentrons.cli analyze --human-json-output - /Users/maxmarrone/Code/Opentrons/opentrons/analyses-snapshot-testing/files/protocols/OT2_S_v3_P300SGen1_None_Gen1PipetteSimple.json

Bisecting with this points to commit 8f2583f as the culprit, and confirms that reverting it, like this PR does, fixes it. But the analyses-snapshot-testing workflows running on this PR show it not fixing it. Something weird is going on there.

I'm going to merge this, trusting my local bisect over our CI environments. I suspect something confusing is happening with the way the workflow chooses the refs to test?

[SyntaxColoring]

Yep, looks like it's fixed now that this is in edge.