feat: add enforce_concatenated_form

[agoose77]

Tip

This PR makes the following changes:

1. Merging categoricals is an error, see Categoricals should merge subject to particular rules #2853
2. _mergemany no longer enforces a backend
3. More bugs in merging are fixed
4. New ak.operations.ak_concatenate.enforce_concatenated_form for enforcing a concatenated form.

dask-awkward needs to be able to compute a concatenation operation but without actually concatenating buffers. @jpivarski and I discussed the various ways we could go about this:

1. Concatenate some typetracer arrays and use enforce_type to produce partitions of the correct type.
2. Concatenate each partition with a length-zero array of the final type.
3. Concatenate some typetracer arrays and introduce a new enforce_form to produce partitions of the correct form.
4. Add a special path to _mergemany that doesn't touch buffers if they have known length-zero.

Each of these solutions has drawbacks. In particular, modifying existing code to satisfy two constraints (perform an optimal merge in convenient amounts of code, and have a length-zero path) is more difficult than having a dedicated function that only handles the "don't touch buffers" case. As such, this PR instead adds a new function that acts like enforce_form but under the conditions that the form was build from ak.concatenate((layout, ...), axis=0) where layout is the content being enforced to form. This allows us to make many simplifying assumptions according to the implementation of _mergemany.

[codecov]

Codecov Report

Merging #2860 (6cab4b0) into main (123fa09) will increase coverage by 0.05%.
The diff coverage is 77.88%.

Additional details and impacted files

Files	Coverage Δ
src/awkward/contents/bitmaskedarray.py	70.81% <100.00%> (+1.53%)	⬆️
src/awkward/contents/bytemaskedarray.py	89.75% <100.00%> (+0.73%)	⬆️
src/awkward/contents/content.py	74.81% <100.00%> (-0.28%)	⬇️
src/awkward/contents/listoffsetarray.py	82.86% <100.00%> (+0.13%)	⬆️
src/awkward/contents/numpyarray.py	91.18% <100.00%> (ø)
src/awkward/contents/regulararray.py	87.43% <100.00%> (+0.18%)	⬆️
src/awkward/contents/unmaskedarray.py	74.60% <100.00%> (+0.39%)	⬆️
src/awkward/contents/listarray.py	91.78% <83.33%> (+1.06%)	⬆️
src/awkward/contents/unionarray.py	85.22% <75.00%> (-0.25%)	⬇️
src/awkward/contents/indexedarray.py	80.95% <82.35%> (+2.05%)	⬆️
... and 2 more

... and 1 file with indirect coverage changes

[jpivarski]

Nice! I noticed all of the typetracer tests that were added to tests/test_0449_merge_many_arrays_in_one_pass.py, too.

We talked about this in our meeting, and the code looks good to me, so this is ready to merge.