Improve decompression threading

[wcampbell0x2a]

A current bottleneck that is that a thread is created per file, instead of per decompressed chunk. This leaves to times where huge files are decompressed with only a single core instead of spreading the chunks that need decompressed into different cores.

This is seen mostly with the following benchmark: https://github.com/wcampbell0x2a/backhand/blob/master/BENCHMARK.md#plexamp-461appimage