Fix fsdp checkpoint strategy

[bcui19]

What does this PR do?

This PR allows us to now use method 2 for activation checkpointing. Basically FSDP recursively dives down to see if there are modules that should have activation checkpointing. However, we shouldn't be trying to wrap 'FullyShardedDataParallel' or 'FlattenedParamsWrappers' modules with the 'CheckpointWrapper'. By trying to wrap these two methods, it leads to a strange assert

Ran with 7b param gpt-2 style models:
run with method 1: https://wandb.ai/mosaic-ml/fsdp-activation-checkpointing/runs/17oacy01
run with method 2: https://wandb.ai/mosaic-ml/fsdp-activation-checkpointing/runs/1f1ij64d

What issue(s) does this change relate to?

Fixing CO-1377

Before submitting

• Have you read the contributor guidelines?
• Is this change a documentation change or typo fix? If so, skip the rest of this checklist.
• Was this change discussed/approved in a GitHub issue first? It is much more likely to be merged if so.
• Did you update any related docs and document your change?
• Did you update any related tests and add any new tests related to your change? (see testing)
• Did you run the tests locally to make sure they pass?
• Did you run pre-commit on your change? (see the pre-commit section of prerequisites)

[abhi-mosaic]

Thank you for the quick fix! I think there is some duplicate code to cleanup, but approving now to unblock.