Support NumPy 2

[cgohlke]

Description

This PR addresses some issues with NumPy 2.0 (see migration guide), which is being released soon:

• Verify that dependencies support numpy 2:

  • matplotlib: 3.8.4
  • tifffile: 2024.4.24
  • lfdfiles: 2024.4.24
  • ptufile: 2024.4.24
  • sdtfile: 2024.4.24
  • czifile: needs rewrite
  • imagecodecs: next release

• The representation of numpy scalars changed. For example, repr(numpy.float64(2.0)) is now 'np.float64(2.0)', was '2.0'. This broke many doctests. The workaround is to enable legacy printing during doctests.

• numpy.array(..., copy=False) now raises an exception if the array can't be copied. The new option copy=None confuses MyPy. The workaround is to replace statements like numpy.array(..., copy=False, ndmin=1) with numpy.atleast_1d(numpy.asarray(...))

• Build wheels against numpy 2 (>=2.0.0rc1) and test with numpy 2 and numpy>=1.23.5,<2.

Release note

Summarize the changes in the code block below to be included in the
release notes:

Support NumPy 2

Checklist

• The pull request title, summary, and description are concise.
• Related issues are linked in the description.
• New dependencies are explained.
• The source code and documentation can be distributed under the MIT license.
• The source code adheres to code standards.
• New classes, functions, and features are thoroughly tested.
• New, user-facing classes, functions, and features are documented.
• New features are covered in tutorials.
• No files other than source code, documentation, and project settings are added to the repository.

[bruno-pannunzio]

Hi @cgohlke, good you already anticipated this changes, as I was not aware of them. It makes sense to start now already implementing the changes to make it compatible. I haven't had much time to see into detail all the changes of Numpy 2 but maybe is something we can discuss on monday's meeting so we all know the impact of this. But I think this can be merged already and then we can see what we are missing