Document the layers of the ctapipe API better

[kosack]

This was noted in discussions with ACADA's SAG-RECO team who are considering using some parts of ctapipe. It was not obvious to them that there are multiple ways to use ctapipe:

• Low-level functions / Components that implement basic algorithms (e.g. hillas_parameters, ImageCleaner, DataWriter)
• Medium-level Components that standardize a complex set of steps like DLx → DLy transforms (e.g. ShowerProcessor)
• High-level Tools that handle configuration, many Components, I/O, provenance, etc, and define command-line applications.
• Transverse functionality, e.g. the instrument and core.metadata modules, maybe the container structure in general?

Showing this "stack" would be useful in the documentation, with examples of how each could be used.

We also noticed that using a Tool outside of a command prompt (i.e. in a python script) is also not well described, though it is somewhat covered in the tool tutorial.

[kosack]

The summary is maybe more that we need a proper architecture section of the docs, with some text or diagrams showing the difference between the various parts:

• algorithms (with the three levels above)
• data model,
• support functionality like instrument, provenance, etc.

May take a bit of thought to do this properly.