I'm not sure if this answers your questions, but...

The code in the generic kernel only understands root/swap on 'a' and 'b' or
both on 'b' if the '*' is specified after the drive spec. Although this is
the way I remember a 4BSD generic kernel doing things 10 years ago
(tradition yah know:-), I don't see any technical reason that it has to
be that way. (I suspect that it was partially because disks aren't usually
labelled properly at this point and you're lucky if 'a' and 'b' are in the
right place, let alone the rest:-)

As for read-only, the root fs always comes up read only for "safety" reasons,
but can be made read/write by updating the mount. For example:

# mount -u /dev/ra0a /

If bootstrapping is refined to the point where a proper disk label with all
partitions is there early on, then I don't see why a generic kernel can't
query for root and swap separately and put them on any partition you like.

Good luck with it, rick