[Question] Macro for type conversion

[Cheukting]

I have been digging around a little bit, and from the tutorial here:
https://marcogorelli.github.io/polars-plugins-tutorial/abs/

We will have to do some type conversion manually like this:

#[polars_expr(output_type_func=same_output_type)]
fn abs_numeric(inputs: &[Series]) -> PolarsResult<Series> {
    let s = &inputs[0];
    match s.dtype() {
        DataType::Int32 => Ok(impl_abs_numeric(s.i32().unwrap()).into_series()),
        DataType::Int64 => Ok(impl_abs_numeric(s.i64().unwrap()).into_series()),
        DataType::Float32 => Ok(impl_abs_numeric(s.f32().unwrap()).into_series()),
        DataType::Float64 => Ok(impl_abs_numeric(s.f64().unwrap()).into_series()),
        dtype => {
            polars_bail!(InvalidOperation:format!("dtype {dtype} not \
            supported for abs_numeric, expected Int32, Int64, Float32, Float64."))
        }
    }
}

I think this will be used commonly enough and wonder if there is already a macro provided in this library that will do this? If not, I am happy to help contribute to one.

I am thinking about one that takes in a numeric series and returns a corresponding ChunkedArray type. Something like:

let ca = to_chunkerd_array!(input[0])?;

Instead of a huge match cause:

let ca = match input.dtype() {
        DataType::Int32 => input.i32()?,
        DataType::Int64 => input.i64()?,
        DataType::Float32 => input.f32()?,
        DataType::Float64 => input.f64()?,
        dtype => {
            polars_bail!(InvalidOperation:format!("dtype {dtype} not \
            supported, expected Int32, Int64, Float32 or Float64."))
        }
    }

CC @MarcoGorelli