DOC: adjust doctests to changes in array repr from numpy 2.0

[neutrinoceros]

A couple non-user facing adjustments to numpy 2.0

example logs

[neutrinoceros]

I don't get why CI is red yet. Even on the failing job it looks like all tests passed. Let me give it another try before I dive into this potential rabbit hole.

[neutrinoceros]

I see GitHub Actions re-released several Python binaries in the last couple days, so let's try again just again the problem is already resolved.

[neutrinoceros]

Alright, new problem: now Windows is failing because it doesn't yet get to install numpy 2. This is caused by the latest version of astropy (6.1.1) pinning numpy<2.0 specifically on windows (and I'm really glad I authored this pin, otherwise I'm not sure I would have found out)
I also know this will be resolved in astropy 6.1.2 (coming soon), and that 6.1.0 didn't have the pin at all. Since I don't think astropy bugs discovered/patched since 6.1.0 actually affect unyt I'm just going to try and exclude 6.1.1 and see how it goes.

[neutrinoceros]

turns out I was just not scanning logs correctly and there was an actual failure on Python 3.9 with minimal supported numpy. Trying to fix this now.