In its actual state, clearance value calculation for an entire map works quite well, but very is slow, especially on
huge maps. I should probabaly try an alternate implementation, compare it to the actual one and then keep what seems to be the fastest.
There is an available source, based on a recursive approach that can be taken as a reference for that alternate implementation.
Updated specs, as of upstream changes