Problem with empty dataframe being returned when querying with only hive partition column.

[allinux]

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

// empty DATAFRAME
LazyFrame::scan_parquet("s3://.../DATE_PARTITION=20221010/*", args)?
	.group_by([col("DATE_PARTITION")])
	.agg([count()])
	.collect();

Log output

No response

Issue description

#11690

When the condition is a dataframe(with columns variable) consisting of only partition columns, df may be empty, so calculating the number of rows in df may result in 0.

fn rg_to_dfs_optionally_par_over_columns(
...
	let mut df = DataFrame::new_no_checks(columns); 
	if let Some(rc) = &row_count {
		df.with_row_count_mut(&rc.name, Some(*previous_row_count + rc.offset));
	}
	materialize_hive_partitions(&mut df, hive_partition_columns);
...
}		
		
fn materialize_hive_partitions(df: &mut DataFrame, hive_partition_columns: Option<&[Series]>) {
    if let Some(hive_columns) = hive_partition_columns {
        let num_rows = df.height();	// 0 

        for s in hive_columns {
            unsafe { df.with_column_unchecked(s.new_from_index(0, num_rows)) };
        }
    }
}

// empty DATAFRAME
LazyFrame::scan_parquet("s3://.../DATE_PARTITION=20221010/*", args)?
	.group_by([col("DATE_PARTITION")])
	.agg([count()])
	.collect();

Expected behavior

.

Installed versions

rustc 1.74.0-beta.2 (9326de8fa 2023-10-13)

[dependencies]
polars = { git = "https://github.com/pola-rs/polars.git", rev = "f40eea6952846f3570382c5b1145e13dd3310949", features = ["async", "lazy", "async", "aws", "parquet"] }

[nameexhaustion]

Thanks for the report - will take a look.