Regression from 0.20.21 -> 0.20.22-rc.1 pl.Expr.list.to_array(n) is throwing polars.exceptions.ComputeError: not all elements have the specified width n

[kszlim]

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

I have a ldf in memory already, this isn't a pure MRE because it seems to rely on memory layout in some way

The following cases seem to fail:

ldf.with_columns(pl.col("some_list_col").list.to_array(3).alias("some_arr_col")).collect()

ldf.collect().with_columns(pl.col("some_list_col").list.to_array(3).alias("some_arr_col"))

But these succeed:

This one is interesting, as rechunk at the DF level seems to solve it, looks like it's a memory layout issue.

ldf.collect().rechunk().with_columns(pl.col("some_list_col").list.to_array(3).alias("some_arr_col"))

ldf.select(pl.col("some_list_col").list.to_array(3).alias("some_arr_col")).collect()

ldf.collect().write_parquet("output.parquet")
df = pl.read_parquet("output.parquet")
df.with_columns(pl.col("some_list_col").list.to_array(3).alias("some_arr_col"))

Log output

`run StackExec` # this happens in the:
`ldf.collect().with_columns(pl.col("some_list_col").list.to_array(3).alias("some_arr_col"))` case

Issue description

When converting a list to array, it seems to have a shape error:
ComputeError: not all elements have the specified width 3

Despite me checking that we have no nulls, all lists are of length 3. It works in a select context too (as opposed to with_columns).

I've gone through the versions and have bisected to this pull request which introduces this regression #15686

Might relate to #16540?

Expected behavior

Should work (and create a new column of array[f64, 3]

Installed versions

--------Version info---------
Polars:               0.20.31
Index type:           UInt32
Platform:             Linux-5.10.217-183.860.x86_64-x86_64-with-glibc2.26
Python:               3.11.7 (main, Dec  5 2023, 22:00:36) [GCC 7.3.1 20180712 (Red Hat 7.3.1-17)]

----Optional dependencies----
adbc_driver_manager:  <not installed>
cloudpickle:          3.0.0
connectorx:           <not installed>
deltalake:            <not installed>
fastexcel:            <not installed>
fsspec:               2024.5.0
gevent:               <not installed>
hvplot:               <not installed>
matplotlib:           3.8.4
nest_asyncio:         1.6.0
numpy:                1.26.4
openpyxl:             <not installed>
pandas:               2.2.2
pyarrow:              16.1.0
pydantic:             2.7.2
pyiceberg:            <not installed>
pyxlsb:               <not installed>
sqlalchemy:           <not installed>
torch:                <not installed>
xlsx2csv:             <not installed>
xlsxwriter:           <not installed>

[kszlim]

@ritchie46 would it be acceptable to revert that PR? It seems like it's spawned a couple bugs.

Seems like #16733 fixes the issue