rolling.agg not returning the correct schema when fill_null is used

[jackaixin]

Checks

• I have checked that this issue has not already been reported.
• I have confirmed this bug exists on the latest version of Polars.

Reproducible example

pl.LazyFrame({
    'a': [1, 2, 3, None, 5]
}).rolling(index_column=pl.int_range(pl.len()).alias('idx'), period='3i').agg(
    pl.col('a'),
    pl.col('a').fill_null(0).alias('a_1')
).collect_schema()

Log output

Schema([('idx', Int64), ('a', List(Int64)), ('a_1', Int64)])

Issue description

The schema of rolling.agg is wrong when fill_null is used.
In the example above, we applied fill_null to pl.col('a') and the schema switched from a list to a scalar.

Expected behavior

I expect the following schema output
Schema([('idx', Int64), ('a', List(Int64)), ('a_1', List(Int64))])

Installed versions

--------Version info---------
Polars:              1.8.2
Index type:          UInt32
Platform:            macOS-15.0-arm64-arm-64bit
Python:              3.11.9 (main, Apr  2 2024, 08:25:04) [Clang 15.0.0 (clang-1500.3.9.4)]

----Optional dependencies----
adbc_driver_manager  1.2.0
altair               5.4.1
cloudpickle          3.0.0
connectorx           0.3.3
deltalake            0.20.1
fastexcel            0.11.6
fsspec               2024.9.0
gevent               24.2.1
great_tables         0.12.0
matplotlib           3.8.2
nest_asyncio         1.6.0
numpy                1.26.2
openpyxl             3.1.5
pandas               2.1.4
pyarrow              17.0.0
pydantic             2.9.2
pyiceberg            0.7.1
sqlalchemy           2.0.35
torch                <not installed>
xlsx2csv             0.8.3
xlsxwriter           3.2.0