Regression from 0.17.15: list.eval fails on data frame with single row

[datenzauberai]

Polars version checks

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

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

Issue description

With the new release our code failed when trying to check if a list contains values from another list. This error only occurs when the data frame contains a single row and the list to compare with has more elements than the list element in the data frame.

Reproducible example

import polars as pl

df = pl.DataFrame(
    {
        "row_number": [0, 1],
        "str_list": [["a", "b"], ["c", "d"]],
    }
)

cmp_list = ["a", "b", "c"] # only a or a, b will not error

# this works (filtering to a single row after list evaluation)
(df
    .with_columns(
        pl.col("str_list").list.eval(
            pl.element().is_in(cmp_list)
        ).alias("comparison")
    ).filter(pl.col("row_number")==0)
)

# this errors, but should yield the same result as above (filtering to a single row before list evaluation)
# InvalidOperationError: output length of `map` (3) must be equal to the input length (2); consider using `apply` instead
(df
    .filter(pl.col("row_number")==0)
    .with_columns(
        pl.col("str_list").list.eval(
            pl.element().is_in(cmp_list)
        ).alias("comparison")
    )
)

Expected behavior

The code should work on a data frame with a single row the same way as on a data frame with multiple rows. This used to work in 0.17.5 using the arr namespace.

Installed versions

--------Version info---------
Polars:      0.18.0
Index type:  UInt32
Platform:    macOS-13.3.1-arm64-arm-64bit
Python:      3.11.3 (main, May 13 2023, 07:40:09) [Clang 14.0.3 (clang-1403.0.22.14.1)]

----Optional dependencies----
numpy:       <not installed>
pandas:      <not installed>
pyarrow:     <not installed>
connectorx:  <not installed>
deltalake:   <not installed>
fsspec:      <not installed>
matplotlib:  <not installed>
xlsx2csv:    <not installed>
xlsxwriter:  <not installed>

[datenzauberai]

Maybe related:

pl.DataFrame({"id": [], "value": []}).groupby("id").agg(pl.col("value").pow(2).mean()) fails on 0.18.0 and not on 0.17.15