ENH: add unyt.testing.assert_equal_units

[neutrinoceros]

In yt-project/yt#3463, it became apparent that in some cases numpy.testing.assert_equal wasn't enough to validate that a unyt_array instance wasn't changed (it doesn't validate that units stayed constant)
This patch implements unyt.testing.assert_equal_units in line with unyt.testing.assert_allclose_units.
I also added unyt_array.is_convertible_to as a helper function.

[neutrinoceros]

making this a draft for now. I'd like to carefully inspect that the docstrings (essentially copy-pasted) still make sense, but I have to run.

[neutrinoceros]

While working on this I found an annoying bug that I think comes from numpy where error messages from array comparison functions (like numpy.testing.assert_array_equal) omit the unit when representing the compared array, leaving a confusing error message in a failing test, e.g.,

E       AssertionError:
E       Not equal to tolerance rtol=1e-07, atol=0
E
E       Mismatched elements: 1 / 3 (33.3%)
E       Max absolute difference: 1.
E       Max relative difference: 0.25
E        x: unyt_array([1, 2, 3])
E        y: unyt_array([1.e-05, 2.e-05, 4.e-05])

I'll remove the draft status when I'm able to confirm the bug isn't in unyt or when I have a fix for it.

[neutrinoceros]

fell down a rabbit hole. I'm now pretty convinced that the interface to fix the issue I mentioned above is implemented as part of NEP 18.

[neutrinoceros]

confirmed that the missing part can be solved by implementing NEP 18. I started doing that in #200. Now I need to study the NEP in greater detail to find out wether the present PR is still relevant in this context (we could probably just "overload" np.testing.assert_array_compare ?).

[neutrinoceros]

I believe #200 actually supersedes this, closing.