Use strategy instead of plugins for DDP

[ejm714]

Changes:

• properly sets up DDP

LightningDeprecationWarning: Passing <pytorch_lightning.plugins.training_type.ddp.DDPPlugin object at 0x7f9b62e37cd0> `strategy` to the `plugins` flag in Trainer has been deprecated in v1.5 and will be removed in v1.7. Use `Trainer(strategy=<pytorch_lightning.plugins.training_type.ddp.DDPPlugin object at 0x7f9b62e37cd0>)` instead.

For context, see PTL PR #12696

Bonus fixes:

• add python extra for mkdocstrings
• typo fixes
• pin protocol buffer version to make tensorboard logger work

[github-actions]

🚀 Deployed on https://deploy-preview-187--silly-keller-664934.netlify.app

[ejm714]

For context, see this github action error and protocol buffers update

[codecov-commenter]

Codecov Report

Merging #187 (e397a12) into master (957f936) will not change coverage.
The diff coverage is n/a.

@@          Coverage Diff           @@
##           master    #187   +/-   ##
======================================
  Coverage    85.2%   85.2%           
======================================
  Files          30      30           
  Lines        1854    1854           
======================================
  Hits         1581    1581           
  Misses        273     273           

Impacted Files	Coverage Δ
zamba/models/config.py	97.1% <ø> (ø)
zamba/models/model_manager.py	84.1% <ø> (ø)