Invalid state DataFrame creatable (columns with different lengths)

[Julian-J-S]

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

df = pl.DataFrame(
    {
        "a": [1, 2],
        "b": [None, None],
    }
)

# this DataFame has unequal column lengths (1 and 0) but CAN be created
df2 = df.select(
    pl.all().drop_nulls().first().name.suffix("_first"),
)

# printing panics
print(df2) # PanicException: The column lengths in the DataFrame are not equal.

# but can still work with it
df2.with_columns(
    pl.all().len()
)
shape: (1, 2)
┌─────────┬─────────┐
│ a_first ┆ b_first │
│ ---     ┆ ---     │
│ u32     ┆ u32     │
╞═════════╪═════════╡
│ 1       ┆ 0       │
└─────────┴─────────┘

Log output

No response

Issue description

It is possible to create an (invalid?) DataFrame with different column lengths.
Should this be possible?

Expected behavior

It should not be possible to create/calculate a DataFrame with different column lengths.

I think the cause for this is probably (almost) always that some functions return a zero-sized DataFrame which is somehow accepted and not broadcasted correctly? (I came across this when looking into #12363 where first/last return 0-size dfs)

Installed versions

0.19.12

[ritchie46]

You deserve a sticker for finding this! :D