PERF-#6696: Use cached dtypes in fillna when possible.

[AndreyPavlenko]

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 PERF: Use cached dtypes in fillna when possible. #6696
• tests added and passing
• module layout described at docs/development/architecture.rst is up-to-date

[dchigarev]

overall looks good, besides the changes related to laziness, let's separate them and discuss in a separate PR

[dchigarev]

why can't we simply fill new dtypes with the find_common_type results?

Suggested change

	dtype = pandas.Series(value).dtype
	if all(
	find_common_type([t, dtype]) == t for t in self._modin_frame.dtypes
	):
	new_dtypes = self._modin_frame.dtypes
	dtype = pandas.Series(value).dtype
	new_dtypes = pandas.Series({col: find_common_type([t, dtype]) for col, t in self._modin_frame.dtypes.items()})

[AndreyPavlenko]

Because the new dtype depends on whether there are NAs in the frame or not. If there no NAs, the dtype is not changed, otherwise, it's changed to the common. We can't analyze the data here, but we can make the assumption, that the dtype will not changed if it's already the common one.

[dchigarev]

same question here