Optimizes map generation. Currently, 16kb were used per chunk for
internal processing geared towards update time and not save time.
Map generation was optimized in a such was that only 8 bytes are
used per chunk, but since we no longer have granular data as before,
occasional lag spikes may occur as much more data needs to get
processed right as we get it.
Unknowingly, the on_chunk_generated event was permitted to be processed
on 2 surfaces instead of the "arena" one, thus producing a "double
pass" effect. Since now, we'll filter based on name, which surface is
getting processed.
