PERF-#7130: Preserve partition lengths in apply_full_axis with keep_partitioning=True

[YarShev]

What do these changes do?

• first commit message and PR title follow format outlined here

  NOTE: If you edit the PR title to match this format, you need to add another commit (even if it's empty) or amend your last commit for the CI job that checks the PR title to pick up the new PR title.

• passes flake8 modin/ asv_bench/benchmarks scripts/doc_checker.py
• passes black --check modin/ asv_bench/benchmarks scripts/doc_checker.py
• signed commit with git commit -s
• Resolves Preserve partition lengths in apply_full_axis with keep_partitioning=True #7130
• tests passing
• module layout described at docs/development/architecture.rst is up-to-date

[anmyachev]

Do we need to check this?

Suggested change

and len(new_partitions.shape) > 1

[YarShev]

Yes, we have to check this. I faced some errors without this check. I think it is because we use np.array([]) and np.array([[]]) in different places to create an empty frame. We should look into this as part of a separate issue. I will file one for this.

[anmyachev]

Can we just this change?

Suggested change

	elif len(new_columns) == 1 and new_partitions.shape[1] == 1:
	kw["column_widths"] = [1]
	elif new_partitions.shape[1] == 1:
	kw["column_widths"] = [len(new_columns)]

[anmyachev]

Same for row_lengths a few lines above.

[YarShev]

This is going to happen in #7133.

[anmyachev]

Note: it (#7131 (comment)) can be done instead of the following changes:

                if (
                    kw["column_widths"] is None
                    and ModinIndex.is_materialized_index(new_columns)
                    and len(new_partitions.shape) > 1
                    and new_partitions.shape[1] == 1
                ):
                    kw["column_widths"] = [len(new_columns)]

This way the diff will be smaller.

[YarShev]

Your suggestion relates to the problem #7133 resolves. We should make changes not by diff size but an issue related.

[anmyachev]

Note: it (#7131 (comment)) can be done instead of the following changes:

                if (
                    kw["column_widths"] is None
                    and ModinIndex.is_materialized_index(new_columns)
                    and len(new_partitions.shape) > 1
                    and new_partitions.shape[1] == 1
                ):
                    kw["column_widths"] = [len(new_columns)]

This way the diff will be smaller.

This block of code is independent of keep_partitioning parameter, are you planning to get rid of the code duplication in #7133?

[YarShev]

I am kind of confused. Your suggestion lies under if not keep_partitioning branch. If we have duplication in #7133, we can get rid of it there.