BUG: groupby behavior changed since 1.4.x

[Yikun]

Pandas version checks

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

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

• I have confirmed this bug exists on the main branch of pandas.

Reproducible Example

# After 1.4.x
>>> import pandas as pd
>>> pdf = pd.DataFrame({"timestamp": [0.0], "car_id": ["A"]})
>>> pdf.groupby("car_id").apply(lambda _: pd.DataFrame({"column": [0.0]})).index
Int64Index([0], dtype='int64')
>>> pdf.groupby("car_id").apply(lambda _: pd.DataFrame({"column": [0.0]}))
   column
0     0.0

Issue Description

groupby behavior changed since 1.4.x

Expected Behavior

# Before 1.4.x
>>> import pandas as pd
>>> pdf = pd.DataFrame({"timestamp": [0.0], "car_id": ["A"]})
>>> pdf.groupby("car_id").apply(lambda _: pd.DataFrame({"column": [0.0]})).index
MultiIndex([('A', 0)],
           names=['car_id', None])
>>> pdf.groupby("car_id").apply(lambda _: pd.DataFrame({"column": [0.0]}))
          column
car_id
A      0     0.0

Installed Versions

1.4.x

[Yikun]

First bad commit: d037ff6

cc @jbrockmendel

[twoertwein]

Potentially a duplicate of #45315

[jbrockmendel]

The bisected commit removed apply_frame_axis0, which was a fastpath that behaved subtly differently from the non-fastpath in ways that we were never really able to consistently debug. The non-fastpath (new and current) behavior is the canonically correct behavior.

[rhshadrach]

In 1.5, you'll be able to specify the group_keys argument to groupby, and apply will adhere to it.

[simonjayhawkins]

Potentially a duplicate of #45315

closing as duplicate and resolved by calling out this breaking change in the 1.4 release notes https://pandas.pydata.org/pandas-docs/stable/whatsnew/v1.4.0.html#groupby-apply-consistent-transform-detection