.implode + .arr.head + .over raises PanicException

[cmdlineluser]

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

While testing out the flatten behaviour in #8965 - I noticed using .arr.head(1) raised an exception.

.arr.slice(0, 1) raises the same exception, but .arr.first() succeeds.

Oddly enough .arr.head(2) also succeeds but returns the same output as .arr.first() which doesn't seem right?

Reproducible example

import polars as pl

df = pl.DataFrame({
    'type': ['water','fire','water','earth']
})


df.select(
   pl.col('type').implode().arr.first().over('type')
)

# shape: (4, 1)
# ┌───────┐
# │ type  │
# │ ---   │
# │ str   │
# ╞═══════╡
# │ water │
# │ fire  │
# │ water │
# │ earth │
# └───────┘

df.select(
   pl.col('type').implode().arr.head(1).over('type')
)

# ComputeError: the length of the window expression did not match that of the group
# > group: ("water")
# > group length: 1
# > output: 'shape: (1,)
# Series: '' [str]
# [
# 	"water"
# ]'
# 
# Error originated in expression: 'col("type").list().slice([0, 1]).over([col("type")])'

Expected behavior

No exception.

Installed versions

--------Version info---------
Polars:      0.17.14
Index type:  UInt32
Python:      3.10.9 (main, Jan 14 2023, 16:56:26)

----Optional dependencies----
numpy:       1.23.5
pandas:      2.0.1
pyarrow:     9.0.0
connectorx:  <not installed>
deltalake:   <not installed>
fsspec:      2022.8.2
matplotlib:  3.5.1
xlsx2csv:    0.8
xlsxwriter:  3.0.8

[ritchie46]

This is correct. This will be more clear with the new mapping strategy argument released.

The default strategy of a window expression is "group_to_idx".

Maybe this makes it more clear:

df = pl.DataFrame({
    'type': ['water','fire','water','earth']
})

df.select(
    pl.all(),
   pl.col('type').over('type', mapping_strategy="join").alias("join"),
   pl.col('type').over('type', mapping_strategy="group_to_rows").alias("g2r"),
   pl.col('type').implode().arr.head(1).over('type', mapping_strategy="join").alias("this cannot be mapped to rows, as it doesn't have enough elements")
)

shape: (4, 4)
┌───────┬────────────────────┬───────┬───────────────────────────────────┐
│ type  ┆ join               ┆ g2r   ┆ this cannot be mapped to rows, a… │
│ ---   ┆ ---                ┆ ---   ┆ ---                               │
│ str   ┆ list[str]          ┆ str   ┆ list[str]                         │
╞═══════╪════════════════════╪═══════╪═══════════════════════════════════╡
│ water ┆ ["water", "water"] ┆ water ┆ ["water"]                         │
│ fire  ┆ ["fire"]           ┆ fire  ┆ ["fire"]                          │
│ water ┆ ["water", "water"] ┆ water ┆ ["water"]                         │
│ earth ┆ ["earth"]          ┆ earth ┆ ["earth"]                         │
└───────┴────────────────────┴───────┴───────────────────────────────────┘