Extending what was previously mentioned in #1000, the inheritance of style variables can perhaps be extended to all other values in the table.

Suggested approach from #1520 would be through a metatable function to lookup fallbacks on-the-fly when something is missing.

Also perhaps there should be a mechanism to override default fallback logic, since it may not always make sense for a.b to inherit from a, and also allow for stuff like b inheriting from a, without being a sub-entry.