They've got the right idea!

Why (may I ask) do you wish to revert to compound constructions? As you can see, for all their initital promise, ultimately they become restrictive.

EDIT
It's only right that I actually respond to the question... Er, not as standard I don't think - but there's always trying to fudge an extra parameter into the spot symbol, then seeing if you can reach down into there with a calculation of a kind. But none of this would be "on the fly", you'd be placing regular (but re-worked) spots, dropping ??? all over the show, then need to commit the pull-calc-push pre-print. If you could live without the lack of responsive feedback and only need these to meet a standard, that'd be something worth further exploration.

In the end it's not something I'll be hacking at as we actively push away denoting any specific dimensional value of elements beyond our scope.
/EDIT

Welcome to the forum BTW! :thumbsup:

Modelling floors as two floors is a good idea besides spot elevation. Boundaries of structural floor are different than that of the finish floor, first of all. On the very same structural floor, there are usually many different finish floors (tiles, parquet, vinyl etc.). Structural floor can stretch beneath interior walls, while finish floor ends at the wall boundary and so on.
I would say that modelling floors as compound floors with all layers inside is a bad practice, since it is not how they're built.