[checkpointio] General Checkpointing of Sharded Optimizers

[Fridge003]

📌 Checklist before creating the PR

• I have created an issue for this PR for traceability
• The title follows the standard format: [doc/gemini/tensor/...]: A concise description
• I have added relevant tags if possible for us to better distinguish different PRs

🚨 Issue number

Link this PR to your issue with words like fixed to automatically close the linked issue upon merge

e.g. fixed #1234, closed #1234, resolved #1234

fixed #3951
fixed #3961

📝 What does this PR do?

Summarize your work here.
if you have any plots/diagrams/screenshots/tables, please attach them here.

As is discussed in #3961, I implemented saving/loading feature of sharded optimizers for GeneralCheckpointIO class.
Broadly speaking, the implementation logic is quite similar to saving/loading sharded models. On top of that, I did a lot of arrangements to adapt the design of sharded models to sharded optimizers. To avoid OOM errors during loading sharded optimizers, I borrowed the implementation of Optimizer.load_state_dict() in source code of pytorch and modified it to our cases. This feature for specific plugins will be developed in the future.

A minor fix: Argument 'variant' and 'prefix' are of same use in the code, so I rename all of the 'variant' to 'prefix' to make our code more readable.

Relevant tests are added under tests/test_checkpoint_io/test_general_checkpoint_io.py.

💥 Checklist before requesting a review

• I have linked my PR to an issue (instruction)
• My issue clearly describes the problem/feature/proposal, with diagrams/charts/table/code if possible
• I have performed a self-review of my code
• I have added thorough tests.
• I have added docstrings for all the functions/methods I implemented

⭐️ Do you enjoy contributing to Colossal-AI?

• 🌝 Yes, I do.
• 🌚 No, I don't.

Tell us more if you don't enjoy contributing to Colossal-AI.