Make tracers non-experimental

[evetion]

In order to make our tracer computation non-experimental:

• Implement subtimestepping (Add subtimestepping for the internal tracer calculation #3014)
• Move the tracer config to its own group
• Document the feature (possibly by @Fati-Mon)
• Validate the implementation (that subtimesteps are required, and compared to Delwaq).
• Consider implementing Trace non-boundary nodes #1969 (or at least how the tables would look like to make it non-breaking in the future).
• Consider implementing time tracers.
• Possibly rename concentration to tracer?

@JohnRushKucharski

[gijsber]

renaming concentration to tracer seems a good idea to me, as it reduces expectations on the functionality and makes clear that you (basically) need a water quality model (like Delwaq) if you want to do substance/concentration modelling