[refactor] Add state_dict to loops

[tchaton]

What does this PR do?

This PR adds the templating logic for loops state. This would be used by Fault Tolerant Logic.

Before submitting

• Was this discussed/approved via a GitHub issue? (not for typos and docs)
• Did you read the contributor guideline, Pull Request section?
• Did you make sure your PR does only one thing, instead of bundling different changes together?
• Did you make sure to update the documentation with your changes? (if necessary)
• Did you write any new necessary tests? (not for typos and docs)
• Did you verify new and existing tests pass locally with your changes?
• Did you update the CHANGELOG? (not for typos, docs, test updates, or internal minor changes/refactorings)

PR review

Anyone in the community is free to review the PR once the tests have passed.
Before you start reviewing make sure you have read Review guidelines. In short, see the following bullet-list:

• Is this pull request ready for review? (if not, please submit in draft mode)
• Check that all items from Before submitting are resolved
• Make sure the title is self-explanatory and the description concisely explains the PR
• Add labels and milestones (and optionally projects) to the PR so it can be classified

Did you have fun?

Make sure you had fun coding 🙃

[codecov]

Codecov Report

Merging #8197 (342853c) into master (3e6f884) will decrease coverage by 5%.
The diff coverage is 96%.

@@           Coverage Diff           @@
##           master   #8197    +/-   ##
=======================================
- Coverage      93%     88%    -5%     
=======================================
  Files         212     212            
  Lines       13679   13695    +16     
=======================================
- Hits        12720   12052   -668     
- Misses        959    1643   +684     

[ananthsub]

do you have examples of what types of state the loop needs to save/load?

[carmocca]

do you have examples of what types of state the loop needs to save/load?

Namely progress (dataclasses) and results (ResultCollection)

trainer
    loops:
        fit_loop:
            epoch_loop:
                batch_loop:
                val_loop:
                    ...
        validate_loop:
            ...
        test_loop:
            ...
------
and each loop has:
a_loop:
    progress
    results

[ananthsub]

could you describe how the trainer will resume the states? at what point is the loop's load_state_dict called?

also does what requirements does this impose on composition of loops?