Change S-2 processor to have harmonize_to_old option

[alexgleith]

Tweaked the harmonize_to_old function so that it handles data without the band dimension.

Changed the s-2 processor so that there are two flags, one for scale_and_offset and another for harmonize_to_old.

Array histogram before:

And after:

[alexgleith]

Any issues with me merging this, @jessjaco?

[jessjaco]

1. I'd prefer making changes to the develop branch if possible. I don't have the bandwidth to track changes across multiple branches.
2. I'm concerned about the change to SimpleLoggingAreaTask. I don't currently use it but it is limiting processing to datasets that have a time variable. What about doing this step in the processor?
3. I think I follow the changes in the harmonization - they're to handle datasets? I'd probably just cast to / from a dataset and recast at the end so there's not duplicate logic

[alexgleith]

I'd prefer making changes to the develop branch if possible. I don't have the bandwidth

I don't want to work of the develop branch yet, as I already have code that works off the main branch. Will refactor, but don't want to yet. I'd prefer to get this merged into main and for us to handle the changes from develop separately. I too don't want to track changes in multiple branches!

I'm concerned about the change to SimpleLoggingAreaTask.

Yeah, good point. I'll have a look at how to incorporate that into the processor.

I'd probably just cast to / from a dataset and recast at the end so there's not duplicate logic

I guess so... I'll look at that too.

[alexgleith]

I think this resolves concerns 2 and 3, @jessjaco.

Can we compromise on 1 please?