At our (Chris Morrison, Eric Bellm, John Swinbank) meeting of 2018-10-30, we discussed this ticket. We concluded that it was possible that we'd be able to address it in the November 2018 sprint (effectively, after
DM-14857 and DM-15588 are done, Chris Morrison is free to chose whether he thinks this or DM-15795 will be more fun to work on first). Given that, I'm pulling it into the sprint, but we can always remove it later.
We also discussed implementation. Chris is keen to take this opportunity to write flags directly into bit-packed fields. I am happy with that approach, as long as the bit-packing machinery doesn't add substantially to the implementation time (this is currently a 6 SP ticket; if writing machinery makes it balloon beyond that, we should stop), and as long as there are appropriate abstractions to shield algorithmic code from having to do bitfield manipulations directly (ie, user-level code should always be able to ask for a flag value by name, not to retrieve a bitfield and have to unpack it).