Group By Transform in mo.ui.dataframe(df) does not return valid Polars code

[henryharbeck]

Describe the bug

Group By Transform in mo.ui.dataframe(df) does not return valid Polars code.

Details in example below. It is pretty easy to see what is going wrong.

Environment

In WASM, so unsure how to run marimo env
Instead, here are the package versions I am using

marimo.__version__: "0.10.8-dev3"
polars.__version__: "1.18.0"

Code to reproduce

import marimo as mo
import polars as pl

df = pl.DataFrame({"group": ["a", "a", "b"], "age": [10, 11, 12]})
mo.ui.dataframe(df)

This Transform

Produces this Polars code

df_next = df
df_next = df_next.group_by(["group"], maintain_order=True).agg([pl.col("age_max").max().alias("age_max_max")])

which raises polars.exceptions.ColumnNotFoundError: age_max

The correct Polars code (with no other stylistic adjustments) would be

df_next = df
df_next = df_next.group_by(["group"], maintain_order=True).agg([pl.col("age").max().alias("age_max")])

[mscolnick]

Fixed by #3349