Move metatensor support into dev branch (crop/pad)

[Nic-Ma]

Description

This PR added MetaTensor support for all the crop & pad transforms.

Status

Ready

Types of changes

• Non-breaking change (fix or new feature that would not break existing functionality).
• Breaking change (fix or new feature that would cause existing functionality to change).
• New tests added to cover the changes.
• Integration tests passed locally by running ./runtests.sh -f -u --net --coverage.
• Quick tests passed locally by running ./runtests.sh --quick --unittests --disttests.
• In-line docstrings updated.
• Documentation updated, tested make html command in the docs/ folder.

[Nic-Ma]

Hi @wyli , @rijobro ,

Thanks for sharing the implementation in the metatensor branch, I refer to the PR #4536 and made the change for Pad transform in this PR, could you please help take a look first then I will dive into more transforms?
The main differences between this PR and the implementation in #4536:

1. The transform can accept tensor, metatensor or numpy, but the computation backend is always tensor.
2. The output of the transform is always tensor or metatensor, I see the typehints in metatensor branch are torch.Tensor for transforms input/output now.
3. The value modification are in-place for the image and metadata, same as our current existing transforms.

Thanks.

[Nic-Ma]

Hi @wyli @rijobro ,

I have integrated all the crop and pad transforms and migrated all the crop & pad unit tests from #4536.
Could you please help review it?
Please note that:

1. Some CI tests can't pass so far because having some other dependencies need to update.
2. I dropped the inverse operation of "crop samples" transforms.

Thanks.

[wyli]

thanks, I put some initial comments inline, please help fix the docs error: https://github.com/Project-MONAI/MONAI/runs/7087792214?check_suite_focus=true

[wyli]

looks good to me