feat(python): rework pl.exclude as a pure selector, allowing other selectors as input

[alexander-beedie]

Closes #13247.

• Implemented pl.exclude as a dedicated selector (it now redirects to the new cs.exclude); adds flexibility, combining all input names/dtypes/selectors into a single/compound selector expression before subtracting that from cs.all().
• Adds expr.meta.is_column(), as it's useful to know if an expression is a simple column or not; this allows for some parsing simplifications/extensions here, and a follow-up PR taking additional advantage is coming shortly.

Note

We can consider retiring pl.exclude in the future as it is now a pure redirect to the more capable cs.exclude.

Example

import polars.selectors as cs
import polars as pl

df = pl.DataFrame({"a": [1], "b": [2], "c": [3]})

df.with_columns(
    pl.exclude(cs.first(), cs.last()) + 100
)
# shape: (1, 3)
# ┌─────┬─────┬─────┐
# │ a   ┆ b   ┆ c   │
# │ --- ┆ --- ┆ --- │
# │ i64 ┆ i64 ┆ i64 │
# ╞═════╪═════╪═════╡
# │ 1   ┆ 102 ┆ 3   │
# └─────┴─────┴─────┘