implode results in extra level of nesting when run within a group_by(...).agg

[jcrist]

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

import polars as pl

df = pl.DataFrame({"x": range(10)})

expr = pl.col("x").implode().alias("res")

# Apply expression without a group_by
no_group_by = df.select(expr)

# Apply expression within a group_by
with_group_by = df.group_by(pl.col("x") > 4).agg(expr)

# The dtype with the group_by is a nested list (List(List(Int64))), which is
# unexpected. For other aggregates the dtype is the same within a group_by as
# it is top-level.
expected_dtype = pl.List(pl.Int64)
assert no_group_by["res"].dtype == expected_dtype
assert with_group_by["res"].dtype == expected_dtype

Log output

keys/aggregates are not partitionable: running default HASH AGGREGATION
Traceback (most recent call last):
  File "/home/jcristharif/Code/ibis/test.py", line 18, in <module>
    assert with_group_by["res"].dtype == expected_dtype
AssertionError

Issue description

When running implode within a group_by(...).agg, an extra level of list nesting is added, which isn't present when run within a select. For other aggregations the dtype when run in a select is the same as the dtype when run within a group_by(...).agg

Expected behavior

The dtype of an aggregation doesn't change when run within a group_by(...).agg.

Installed versions

--------Version info---------
Polars:               0.20.31
Index type:           UInt32
Platform:             Linux-6.5.0-35-generic-x86_64-with-glibc2.38
Python:               3.10.13 | packaged by conda-forge | (main, Oct 26 2023, 18:07:37) [GCC 12.3.0]

----Optional dependencies----
adbc_driver_manager:  <not installed>
cloudpickle:          2.2.1
connectorx:           <not installed>
deltalake:            0.16.0
fastexcel:            <not installed>
fsspec:               2023.12.2
gevent:               <not installed>
hvplot:               <not installed>
matplotlib:           3.8.0
nest_asyncio:         1.5.8
numpy:                1.23.5
openpyxl:             <not installed>
pandas:               1.5.3
pyarrow:              16.1.0
pydantic:             2.4.2
pyiceberg:            <not installed>
pyxlsb:               <not installed>
sqlalchemy:           1.4.49
torch:                <not installed>
xlsx2csv:             <not installed>
xlsxwriter:           <not installed>

[cmdlineluser]

The rationale is explained in #6487 ^{(.implode() was previously known as .list())}

• .implode() returns list(list()) instead of list() #10042

[jcrist]

Ah, thanks. I still find the behavior surprising, but since it's intended then I'll close.