Updated numpy to 2.0.2 and other packages to the latest

[tkakar]

Fixes Issue 362

[NickAkhmetov]

Does pandas also need to be updated? It looks like there's a new major version out as of last year: https://pandas.pydata.org/docs/whatsnew/index.html

CC @keller-mark

[NickAkhmetov]

Actually, after taking a closer look at the changelog, it does seem like a version bump is necessary to maintain interoperability between pandas/numpy: https://pandas.pydata.org/docs/whatsnew/v2.2.2.html#pandas-2-2-2-is-now-compatible-with-numpy-2-0

[tkakar]

@NickAkhmetov thanks for reviewing. I can change it, but there are tests for Python3.8 and Python3.12, so pandas has to be compatible with Python3.8.
Out of curiosity, wouldn't the latest package be downloaded based on the python version?
I have not changed the others' either, as the restrictions seemed fine. can bump those as well (where possible)

[NickAkhmetov]

It would be able to resolve to the latest compatible pandas version with the existing approach in a greenfield project; however, if we mark Vitessce as being compatible with older versions of pandas, it could cause downstream issues for users.

For example, if a user's project has a dependency on pandas<2 and they installed vitessce without the change to pandas>=2.0.2, then the latest version of pandas would not be installed and unexpected runtime errors could arise from whatever incompatibilities pandas<2.0.2 has with numpy>2. By setting a stricter range, we can preempt those errors by having the package manager tell the user "hey, this other package also needs to be updated for things to work as expected."

[tkakar]

Thanks for the explanation. makes sense

[NickAkhmetov]

Actually - on further reflection, I was overthinking it. If the numpy version is >1.21.1, we can actually use the previous pandas version number, since the package manager should be able to resolve the appropriate pairing of numpy/pandas versions that should be used in a given circumstance. Apologies for the confusion 😅

[NickAkhmetov]

Looks good to me! Some non-blocking questions:

• Should this be a minor version bump instead of a patch, since this includes some major dependency version updates?
• This is more of a matter of personal curiosity - is there something specific about the ome-zarr version we're pinned to? There appears to be a new release as of a few months ago, maybe we can set the ome-zarr to >=0.8.3?

[tkakar]

@NickAkhmetov thanks for reviewing. for ome-zarr, I did get errors when I bumped the version, so that could be a reason?
regarding the minor version bump, I agree, as there could be potential consequences of these changes

[tkakar]

@NickAkhmetov I bumped the minor version but tests were failing, reverting it back is still failing the tests (docs build) though the code is the same as the previous successful run.

[tkakar]

@NickAkhmetov numpy/numpy#26710
this might be an issue as python3.12 and docs build workflow fail