PERF-#7299: Avoid using synchronize_labels for combine function

[anmyachev]

What do these changes do?

Perf gain: ~15% against main branch with Ray 8 cores.

import modin.pandas as pd
from modin.utils import execute
import numpy as np

from time import time

df1 = pd.DataFrame(np.random.randint(1_000_000, size=(10_000, 100)))
df2 = pd.DataFrame(np.random.randint(1_000_000, size=(1_000_000, 100)))

for _ in range(5):
    start = time()
    df2._query_compiler._modin_frame._deferred_index = True
    res = df1.merge(df2, on=3)
    execute(res)
    print(f"merge time: {time()-start}")

• 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 Avoid using synchronize_labels for combine function #7299
• tests added and passing
• module layout described at docs/development/architecture.rst is up-to-date

[anmyachev]

A further improvement could be to get rid of the materialization of indexes in the main process. However this also happens in _propagate_index_objs function now.

[devin-petersohn]

This is great @anmyachev!

I think a good medium term goal would be to have support for a fully lazy index object, adding it as a first class citizen to the query compiler (and adding index.py to modin/pandas.

LGTM

[YarShev]

This is great @anmyachev!

I think a good medium term goal would be to have support for a fully lazy index object, adding it as a first class citizen to the query compiler (and adding index.py to modin/pandas.

LGTM

@dchigarev, @anmyachev, to what extent does our current ModinIndex perform this task?