SeriesGroupBy.value_counts() should return `Series` not `DataFrame`

[Dr-Irv]

Describe the bug
From discussion here: #203 (comment)

To Reproduce

import pandas as pd

df = pd.DataFrame(
    {
        "Animal": ["Falcon", "Falcon", "Parrot", "Parrot"],
        "Max Speed": [380, 370, 24, 26],
    }
)
c: pd.Series = df.groupby("Animal")["Max Speed"].value_counts()
print(type(c))  # <class 'pandas.core.series.Series'>
print(c.index)
print(c)

mypy and pyright report that there is a mismatch in the result.

In pandas source, SeriesGroupBy.value_counts() always returns Series . Need to make that change in typing stubs.

pandas-stubs/pandas-stubs/core/groupby/generic.pyi

Line 75 in 44f804c

) -> DataFrame: ...

But should copy overloads from DataFrameGroupBy for value_counts() to handle normalize argument correctly.