DEP: Make np.insert and np.delete on 0d arrays with an axis an error

[eric-wieser]

Before this change, the following code worked:

>>> some_0d = np.array(1)
>>> np.insert(some_0d, "some nonsense", 10, axis=0)
array(10)
>>> np.insert(some_0d, "some nonsense", 42, axis="some nonsense")
array(42)

Now these raise AxisError and TypeError, respectively.

delete is exactly the same.

[eric-wieser]

Googling for in the future the special handling of scalars will be removed gives no results other than clones of numpy, so it looks like no one ever ran into this deprecation anyway.

[seberg]

Probably, I do not feel like bikeshedding on the release notes right now, so going to put this in when the tests are done. Thanks Eric... should have done this years long ago.

[eric-wieser]

Perhaps worth leaving for a few days for someone with release note opinions like @charris or @rgommers to chime in. It's not like anything is blocked by this PR.

[eric-wieser]

Since the other PRs are merged and have very similar structures of release note, may as well merge this one too. Release notes can always be cleaned up at release time.