Feature: migrate to numpy v2.0.0

[mrava87]

This PR aims to allow users to use pylops with numpy>=2.0.0.

Currently we need to make the following work-arounds:

• disable tests for spgl1 until new release of spgl1 library
• disable tests for DTCWT until new release of dtcwt library
• disable tests for TorchOperator in darwin os (currently torch does give this error RuntimeError: Numpy is not available)

[mrava87]

@cako can you please quickly review this one, so we can make a new release that works with numpy v2

[cako]

Approving, LGTM. I added a couple of changes to the testing logic to reduce the diff, and I also added a note in the documentation about dtcwt and spgl1 not working with numpy2

[cako]

@mrava87 Apparently dtcwt has been archived. The new version lives in: https://github.com/xir4n/dtcwt

We might want to support that fork instead.

EDIT: Just saw that the fork is the one being used in PyPI. Nevermind!

[mrava87]

@mrava87 Apparently dtcwt has been archived. The new version lives in: https://github.com/xir4n/dtcwt

We might want to support that fork instead.

EDIT: Just saw that the fork is the one being used in PyPI. Nevermind!

I saw this... it 'happened' as soon as I opened an Issue about numpy2 (rjw57/dtcwt#149).... now, I cannot reply anymore to this Issue thread and the fork does not have Issues (since it is a fork), so we seem to have lost all ways to communicate. I suggest that we keep this for now, it would not affect people using pylops, and consider ditching dtcwt all together in a few months if the library does not update to support numpy2. What do you think?

[cako]

I don't mind keeping it. If it becomes a pain to maintain, we can remove it in the future.