Make ifo_set_type a bitset - Follow-up from "Cherry-Pick Non-Science Changes"

The following discussion from !25 (merged) should be addressed:

• @andrewmichael.gozzard started a discussion: (+3 comments)

  Probably not for this MR, but I have mentioned before and want to make note that I think this is really unreadable and is begging for errors. The empty set is a perfectly valid set that we may want to be able to represent. If we insist on not having any arrays contain it we should offset the indexing of those arrays then and no sooner.

  This will basically mean going through and finding everywhere this combo bitset is a thing now and moving the offset.

  Using the offset in the at-rest encoding means that this bit of logic for interpreting it is highly coupled with the data, and will lead to lots of duplication. We should maybe consider helper functions to centralize this encoding logic.