Add argument `value_drop_null` to `melt()`

[mgirlich]

Description

It is quite common to remove rows that only contain null values in the newly created value column after melting. It would be nice to add an argument value_drop_null to do this.

Example

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

# defaults to False
melted_keep = df.melt("a", value_drop_null=False)
# ┌─────┬──────────┬───────┐
# │ a   ┆ variable ┆ value │
# │ --- ┆ ---      ┆ ---   │
# │ str ┆ str      ┆ i64   │
# ╞═════╪══════════╪═══════╡
# │ x   ┆ b        ┆ 1     │
# │ y   ┆ b        ┆ null  │
# └─────┴──────────┴───────┘

melted_keep = df.melt("a", value_drop_null=True)
# ┌─────┬──────────┬───────┐
# │ a   ┆ variable ┆ value │
# │ --- ┆ ---      ┆ ---   │
# │ str ┆ str      ┆ i64   │
# ╞═════╪══════════╪═══════╡
# │ x   ┆ b        ┆ 1     │
# └─────┴──────────┴───────┘

[avimallu]

Broadly a duplicate of https://github.com/pola-rs/polars/issues/8903.

I think this feature will also be better addressed by Polar's focus on composability of functions (i.e. using a drop_nulls()) over adding a specialized argument over it to keep the API surface small. There's little performance improvement or typing reduction for this argument.