Speed up CI

[daavoo]

From 2h 2m 44s to 21m 48s.

Use pip cache.
Run linters on a single python version.
Decouple ML frameworks tests and run on them a single python version.

Closes #402

[codecov-commenter]

Codecov Report

Base: 96.75% // Head: 96.70% // Decreases project coverage by -0.05% ⚠️

Coverage data is based on head (f336d2d) compared to base (42eff97).
Patch coverage: 100.00% of modified lines in pull request are covered.

Additional details and impacted files

@@            Coverage Diff             @@
##             main     #411      +/-   ##
==========================================
- Coverage   96.75%   96.70%   -0.06%     
==========================================
  Files          37       37              
  Lines        2189     2186       -3     
  Branches      206      181      -25     
==========================================
- Hits         2118     2114       -4     
  Misses         41       41              
- Partials       30       31       +1     

Impacted Files	Coverage Δ
tests/test_frameworks/test_catalyst.py	95.55% <ø> (ø)
tests/test_frameworks/test_fastai.py	100.00% <ø> (ø)
tests/test_frameworks/test_huggingface.py	97.82% <ø> (ø)
tests/test_frameworks/test_keras.py	100.00% <ø> (ø)
tests/test_frameworks/test_lgbm.py	100.00% <ø> (ø)
tests/test_frameworks/test_lightning.py	97.05% <ø> (ø)
tests/test_frameworks/test_xgboost.py	100.00% <ø> (ø)
tests/test_main.py	100.00% <100.00%> (ø)
src/dvclive/dvc.py	90.29% <0.00%> (-1.14%)	⬇️

Help us with your feedback. Take ten seconds to tell us how you rate us. Have a feature suggestion? Share it here.

☔ View full report at Codecov.
📢 Do you have feedback about the report comment? Let us know in this issue.