BUG: Fix numeric_only ignored with list of functions in DataFrame.agg and GroupBy.agg (#49352)

[ali-aqib]

Closes #49352

When numeric_only=True is used with a list of aggregation functions in DataFrame.agg() or GroupBy.agg(), non-numeric columns cause a TypeError because filtering isn't applied before the functions run.

This PR filter to numeric columns before applying the list of functions when numeric_only=True is specified.

---

Changes-

• Modified NDFrameApply and GroupByApply in pandas/core/apply.py to filter numeric columns before aggregation
• Added tests in test_frame_apply_numeric_only.py and test_aggregate_numeric_only.py
• Updated whatsnew for v3.0.0

---

Example

df = pd.DataFrame({
    'key': ['A', 'B', 'A', 'B'],
    'num': [1, 2, 3, 4],
    'text': ['a', 'b', 'c', 'd']
})

# Now works with numeric_only = True
result1 = df.agg(['sum', 'mean'], numeric_only=True) 
result2 = df.groupby('key').agg(['sum', 'mean'], numeric_only=True) 

---

[rhshadrach]

Thanks for the PR. The text in the OP here is very long and appears to be AI generated. It also contains a lot of information that is entirely redundant. Please do not do this - it is mostly noise and very little signal. A short summary of the PR and any additional details or context that are not found elsewhere would be appropriate.

[ali-aqib]

Thanks for the feedback
I've updated the description to be more concise.

[rhshadrach]

Thanks for the PR!

[rhshadrach]

Only two of the five paths in DataFrameGroupBy.aggregate hits here, thus introducing inconsistencies. To accept a PR, we would need to do this consistently through the op. Take a look at the code in pandas.core.groupby.generic.

[ali-aqib]

You're right I need to ensure the filtering is applied consistently across all code paths in DataFrameGroupBy.aggregate.
I'll update the PR.