Is your feature request related to a problem or challenge? Please describe what you are trying to do.
I am trying to arithmetically create date / time columns. In my case I want to add some duration (7 days) to a Date32 column

Describe the solution you'd like

I would like temporal kernels that allowed operations such as:

TimestampNanosecondArray + Duration --> TimestampNanosecondArray
Date32Array + Duration --> Date32Array
etc.

I would like to be able to do something like

    let date_strings = vec![
        Some("2020-01-01"),
        Some("2020-01-02"),
        Some("2020-01-03"),
        None,
        Some("2020-01-04"),
    ];
    let arr_date32 = arrow::compute::cast(&arr_string, &DataType::Date32).expect("casting to date");

    // note there is no such thing as DurationArray at this time
    let offset_seconds = DurationArray::from_days(vec![1, 2, 3, 4, 5]);
    let arr_date32_offset = compute::kernels::add(arr_date32, offset_seconds)

    // expect the output is 
    // 2021-01-02
    // 2021-01-04
    // 2021-01-06
    // None
    // 2021-01-08

Describe alternatives you've considered
It might be nice to extend the arithmetic kernels to take Arc<dyn Array> rather than the PrimitiveArray as they do now and do the casting / calling specific implementations directly.

Additional context
Add any other context or screenshots about the feature request here.