DataFrame.__pow__ fails for (series, column) inputs

[mcrumiller]

Issue Description

When pow() is used in a select, if the first argument is an external Series and the second is a column, an invalid ColumnNotFoundError is raised.

Noticed as a result of looking into #17760.

import polars as pl

df = pl.DataFrame({"a": [1, 1, 2]})
s = pl.Series("b", [1, 1, 1])
df.select(s ** pl.col("a"))

polars.exceptions.ColumnNotFoundError: a

Installed versions

main

[mcrumiller]

The real error is in Series * expr failing if expr is a column:

pl.Series() * pl.col("a")   # <Expr ['[(Series) * (col("a"))]'] at 0x7F19A0B5CE60>
pl.Series() ** pl.col("a")  # polars.exceptions.ColumnNotFoundError: a