Derived variables for forecast data

[joeloskarsson]

It would be highly useful to also be able to derive variables for forecast datasets. The difference with those is that they don't just have a time dimension, but an analysis time and a lead time dimension.

I don't think there is anything inherent in the way derived variables are computed that prevent doing it for forecast datasets, it is just that the current derived variables do not allow for this (analysis time, lead time) pair of dimensions as input (and correctly using these).

[joeloskarsson]

I implemented one way to allow for this here: sadamov@b1901a1 It is quite easy thanks to the flexible design of derived variables are computed 😄

It would be great if someone wanted to take that and turn it into a full-fledged PR. It might need some cleanup + it would be great to get @ealerskans's input on it.