Allow group_by on lists of strings

[ByteNybbler]

Description

The following code currently panics with the following verbose output:

use polars::prelude::*;

fn main() {
    let num_rows = 1_100;

    let df = df![
        "lists" => vec![Series::new("inner", ["str"; 1]); num_rows],
    ].unwrap();

    df.lazy()
        .group_by([col("lists")])
        .agg([col("lists").alias("agg")])
        .collect()
        .unwrap();
}

run GroupbyExec
keys/aggregates are not partitionable: running default HASH AGGREGATION
thread 'main' panicked at src/issuex.rs:14:10:
called `Result::unwrap()` on an `Err` value: ComputeError(ErrString("grouping on list type is only allowed if the inner type is numeric"))
note: run with `RUST_BACKTRACE=1` environment variable to display a backtrace

It would be nice if lists of strings could be used as hash join keys.

[cmdlineluser]

If it helps: You can cast to categorical.

df = pl.DataFrame({"lists": [["A"], ["B"], ["A"]], "val": [1, 2, 3]})

df.group_by(pl.col("lists").cast(pl.List(pl.Categorical))).all()
# shape: (2, 2)
# ┌───────────┬───────────┐
# │ lists     ┆ val       │
# │ ---       ┆ ---       │
# │ list[cat] ┆ list[i64] │
# ╞═══════════╪═══════════╡
# │ ["A"]     ┆ [1, 3]    │
# │ ["B"]     ┆ [2]       │
# └───────────┴───────────┘

[nardi]

I'm looking for this feature as well (and similarly, joins on list columns). Is there a reason it is not implemented? It seems that there is hashing functionality implemented, so it should be possible to simply hash the list column and join on that, right?

[cmdlineluser]

Closed by #15540