add support for / switch default dbs to zstd

[dpark01]

It's probably about time to make sure we add support for zstd throughout our various bits of code for databases (such as read_utils.py extract_tarball, among other places) and flowcell tarballs so that we can start offering up some of our large databases and sequencing runs as .tar.zst files. Aside from the fact that folks seem to be moving that way in general, including the author of lz4 himself, I've found it pretty easy and fast to use.