fix: support option-in-record for fill_none

[agoose77]

Fixes #2064 (see issue for details). I feel like we might have touched upon this before, so keen for comments.

[codecov]

Codecov Report

Merging #2065 (b34431a) into main (8d7a2c3) will increase coverage by 0.00%.
The diff coverage is 100.00%.

Additional details and impacted files

Impacted Files	Coverage Δ
src/awkward/operations/ak_fill_none.py	95.00% <100.00%> (+2.31%)	⬆️
src/awkward/contents/bytemaskedarray.py	88.80% <0.00%> (+0.25%)	⬆️

[jpivarski]

Some operations that used to act on data within records have been stopped at the RecordArray border, but I see no reason why this one should be. As a user, I would expect to be able to say ak.fill_none on an array containing records and have the missing values in the records get filled in.

It has to respect the axis parameter, but it looks like it does.

I'd say this is ready to merge.

[agoose77]

@jpivarski let me just test one additional case! I'm sorry for pinging you and then pausing the merge; I noticed something peculiar.

[agoose77]

OK, the peculiarity I noticed was if the record contains options, and the record itself is wrapped in an option. This PR does what is_none does; the first wins.