ColumnNotFoundError occurs when ORDER BY references unaliased column in SQL query with aggregation

[BTatlock]

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

pl.DataFrame({"a": [1, 3, 2]}).sql("SELECT a b FROM self GROUP BY a ORDER BY a")

Log output

polars.exceptions.ColumnNotFoundError: "a" not found

Issue description

The column alias is expected by the ORDER BY query, meaning the following query works:

pl.DataFrame({"a": [1, 3, 2]}).sql("SELECT a b FROM self GROUP BY a ORDER BY b")

The unexpected behaviour is also seen when using fully qualified identifiers:

pl.DataFrame({"a": [1, 3, 2]}).sql("SELECT self.a b FROM self GROUP BY self.a ORDER BY self.a")

Expected behavior

Should be equivalent to

pl.DataFrame({"a": [1, 3, 2]}).sql("SELECT a FROM self GROUP BY a ORDER BY a").rename({"a": "b"})

Installed versions

--------Version info---------
Polars:              1.17.1