Improve performance

[bodgit]

Currently, every time a file within the archive is opened, a new copy of the decompressed stream is created and then read until the beginning of the file in question is reached. This means there is a performance hit that gets worse as you descend into the archive. To read the 10th file, you have to read and discard the first 9 files in the archive, to read the 100th file, you have to read past 99 files in the archive, etc.

A performance improvement would be to keep the decompressed stream reader around for any future files that are further along in the archive, (you can't go backwards).