Fix packaging imports in version comparison logic

[nkaenzig]

Fixes #8349

Description

The current behaviour is that pkging, has_ver = optional_import("packaging.Version") always returns has_ver=False because the import always fails (the Version class is exposed by the packaging.version submodule).

This issue previously didn't surface, because when the import fails, it would just continue to use the fallback logic. However, there seem to be more hidden and more severe implications, which ultimately led me to discovering this particular bug: Function like floor_divide() in monai.transforms that check the module version using this logic are called many times in common ML dataloading workflows. The failed imports somehow can lead to OOM errors and the main process being killed (see #8348). Maybe when optional_import fails to import a module, the lazy exceptions somehow stack up in memory when this function is called many times in a short time period?

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.

[KumoLiu]

Thanks for the fix, LGTM.

[KumoLiu]

/build