Feat: raise AttributeError for public Array attributes

[agoose77]

Fixes #1511

As discussed in the issue, the new behaviour is:

• existing public class attributes can be assigned to (though the attribute itself may not be writable, e.g. with a read-only data descriptor)
• private attributes can be assigned to, and will never attempt to set a field
• fields can never be set

We might have some hidden issues if users employ underscore-prefixed fields. Right now, setting these attributes will write to the ak.Array object. However, this shadowing is already accepted when we read from an attribute - class attributes take precedence over fields, just as they do in this PR for writing. Are we happy with this?

[codecov]

Codecov Report

Merging #1573 (284fa23) into main (9e17f29) will decrease coverage by 0.01%.
The diff coverage is 35.13%.

Impacted Files	Coverage Δ
src/awkward/_v2/_connect/cuda/__init__.py	0.00% <0.00%> (ø)
src/awkward/_v2/_connect/numexpr.py	88.40% <0.00%> (ø)
src/awkward/_v2/_connect/pyarrow.py	88.46% <0.00%> (ø)
src/awkward/_v2/contents/bytemaskedarray.py	88.82% <0.00%> (ø)
src/awkward/_v2/contents/indexedarray.py	73.83% <0.00%> (ø)
src/awkward/_v2/contents/indexedoptionarray.py	89.14% <0.00%> (ø)
src/awkward/_v2/contents/listoffsetarray.py	81.85% <0.00%> (ø)
src/awkward/_v2/contents/unionarray.py	86.27% <0.00%> (ø)
src/awkward/_v2/numba.py	93.47% <0.00%> (ø)
src/awkward/_v2/operations/ak_from_avro_file.py	66.66% <0.00%> (ø)
... and 6 more

[jpivarski]

This is good; I'd like to merge it, but I'm going to update the branch to main (after #1576 is done) before doing so.

[agoose77]

I might change ... in dir(type(self)) to hasattr(self.__class__, ...) for simplicity, but I'll do it at a later date.

[jpivarski]

Good point. Or maybe hasattr(type(self)) because if there's a public function that reveals a dunder-attribute, I think it should be preferred.