incorporating support for federated evaluation

[kta-intel]

Request: #788

This PR adds an additional workspace template for federated evaluation using torch_cnn_mnist as a base

Changes made to plan/plan.yaml to set rounds_to_train to 1 and set aggregator_model_validation as the only task to be executed

Changes made to src/pt_cnn.py to remove optimizer initialization and loss function, which are no longer needed for a single inference run

Added .rst file to docs to introduce federated eval and walk through example (Note: landing page TBD)

[psfoley]

@kta-intel the federated evaluation documentation you've added here is excellent. Can you confirm that the page is indexed correctly (i.e. shown on the side pane when you build locally)? If so this is ready to merge.

[kta-intel]

@kta-intel the federated evaluation documentation you've added here is excellent. Can you confirm that the page is indexed correctly (i.e. shown on the side pane when you build locally)? If so this is ready to merge.

Great catch! I added it under the manual tab. Let me know if this looks good or if you think it should be added elsewhere instead. It builds locally without issue

[psfoley]

Thanks for the quick turnaround Kevin. manual is a good spot for the documentation. Only items remaining are 3 minor linting errors. Once those are fixed this can be merged.