Move literal group_by keys out of the grouping

[orlp]

If a group_by key is a literal it could be removed from the grouping. For example, before:

>>> df = pl.DataFrame({"x": [1, 2, 3], "g": [1, 1, 2]})
>>> df.group_by([pl.lit(1), "g"]).agg(pl.col.x.sum())
shape: (2, 3)
┌─────────┬─────┬─────┐
│ literal ┆ g   ┆ x   │
│ ---     ┆ --- ┆ --- │
│ i32     ┆ i64 ┆ i64 │
╞═════════╪═════╪═════╡
│ 1       ┆ 1   ┆ 3   │
│ 1       ┆ 2   ┆ 3   │
└─────────┴─────┴─────┘

After:

>>> df = pl.DataFrame({"x": [1, 2, 3], "g": [1, 1, 2]})
>>> df.group_by("g").agg(pl.col.x.sum()).select(pl.lit(1), pl.col.g, pl.col.x)
shape: (2, 3)
┌─────────┬─────┬─────┐
│ literal ┆ g   ┆ x   │
│ ---     ┆ --- ┆ --- │
│ i32     ┆ i64 ┆ i64 │
╞═════════╪═════╪═════╡
│ 1       ┆ 1   ┆ 3   │
│ 1       ┆ 2   ┆ 3   │
└─────────┴─────┴─────┘

In the event that the literal is the only grouping key, the entire .agg should be rewritten to a select.