Document how to do stage and load as an analogy to ophyd's stage_sigs

[coretl]

Q: Does ophyd-async have a concept similar to stage_sigs in ophyd?
A: No, because one of the biggest slowdowns in our current (non-bluesky) system is "get things ready for a scan", then "put it back to how it was", over and over again. instead we have the concept of load and save plans that could be run in stage and unstage, so that each plan can ready itself for the thing it is about to do, but doesn't have to know about the state to return the beamline to when it's done. We would then tell the level above (like queueserver) what state to return the beamline to when it's finished running it's queue of plans or if one of them fails.
Note that load and save file formats will be changing file formats shortly in preparation for the 1.0 release, so I would hold off on using them for a few weeks.

This needs documenting for V1

Acceptance Criteria

• There are docs on this