BUG: groupby(sort=True) may produce unsorted results with range-partitioning implementation

[dchigarev]

Modin version checks

• I have checked that this issue has not already been reported.

• I have confirmed this bug exists on the latest released version of Modin.

• I have confirmed this bug exists on the main branch of Modin. (In order to do this you can follow this guide.)

Reproducible Example

import numpy as np
import modin.pandas as pd
import pandas

from modin.pandas.test.utils import df_equals

import modin.config as cfg
cfg.IsDebug.put(True)
cfg.NPartitions.put(4)
cfg.RangePartitioningGroupby.put(True)

np.random.seed(214)

data = {
    "a": ["a", "b", "c", "d", "e", "b", "g", "a"] * 32,
    "b": [1, 2, 3, 4] * 64,
    "c": range(256),
    "d": range(256),
    "e": ["x", "y"] * 128,
}

filter = lambda row: (~row["a"].isin(["a", "e"]) & ~row["b"].isin([4]))

md_df, pd_df = pd.DataFrame(data), pandas.DataFrame(data)
md_df = md_df[filter]
pd_df = pd_df[filter]

md_res = md_df.groupby(["a", "e"]).sum()
pd_res = pd_df.groupby(["a", "e"]).sum()
df_equals(md_res, pd_res)
# MultiIndex level [0] values are different (100.0 %)
# [left]:  Index(['c', 'g', 'b'], dtype='object', name='a')
# [right]: Index(['b', 'c', 'g'], dtype='object', name='a')
# At positional index 0, first diff: c != b

Issue Description

The modin's result is unsorted. This seems to be only relevant for multi-column groupby

Expected Behavior

should be sorted

Error Logs

Replace this line with the error backtrace (if applicable).

Installed Versions

Replace this line with the output of pd.show_versions()