Add a cli tool to test nn training functionality

[thodkatz]

This PR adds a command line tool to test the functionality of the neural network training. I have included a README.md for its usage.

I wasn't really sure where to add it, I have created a seperate dir nn_training_cli, under the tiktorch/dev.

[codecov]

Codecov Report

All modified and coverable lines are covered by tests ✅

Project coverage is 62.40%. Comparing base (15a0225) to head (2f61bce).

Additional details and impacted files

@@           Coverage Diff           @@
##             main     #233   +/-   ##
=======================================
  Coverage   62.40%   62.40%           
=======================================
  Files          47       47           
  Lines        2878     2878           
=======================================
  Hits         1796     1796           
  Misses       1082     1082           

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

[thodkatz]

To test it on a gpu, change the device attribute of tiktorch/dev/nn_training_cli/pytorch3d_unet_config.yaml from device: cpu to device: cuda.

Related issue #234, regarding silicon compatibility.

[k-dominik]

Cool! MPS currently is still lacking support for certain operations that prevent it from running in 3D (Def. MaxPool3D is still missing). 2D should work by setting device: mps I'll try :)