Chunks are just meta data, based on the chosen shard key that logically divide the sharded data up into ranges, they do not dictate anything in terms of the storage engine of MongoDB (see here for more information there:
http://www.10gen.com/presentations/mongosv-2011/mongodb-storage-engine-bit-by-bit).
The maximum chunk size is used by the mongos to judge when to request splits, how many split points come back and then how much data is migrated when a chunk moves. There are no side-effects about non-full chunks, besides the moves and such going more quickly because there is less data. In fact one of the ways we try to make the max key chunk migration happen more quickly is to essentially make it a zero data chunk.
Adam