fallback for some level corrupt in wsi

[winglet0996]

Hi, in this patch, line 214 to 231, provides a fallback of reading other level in some case that some level at the pyramid of wsi has corrupted.
Example slide for test: HANCOCK, PrimaryTumor_HE_232

[guillaumejaume]

Thanks for the patch @winglet0996! Could you limit the PR to the read_region patch without including docstrings + _initialized changes.

Let me know if unclear
Best, Guillaume

[winglet0996]

Thanks for your reply, I think the other messy changes stem from my local TRIDENT version is behind the current release and I've updated them for your review.
BTW, are you interested in a multi-GPU feature as in PR #181? It's an old PR and also has some conflicts now...

[guillaumejaume]

Thanks @winglet0996!

Yes, happy to integrate multi GPU support too. I never had the chance to look at the code in details. Ping me when you have the conflicts merged. I think many users will be happy to use multi-gpu inference on large slides.

[winglet0996]

Sounds great.
You may go ahead and merge this and your latest updates (like czi reader) to main first.​ I’ll then resolve the multi-GPU PR conflicts against the newest main branch.
Just let me know when things are ready, thank you.