think about distributing zipfile databases with different scaled values

[ctb]

the unicity work (more on that later) suggests that some genomes need different scaled values to be distinguishable from each other within a reference database. we could calculate "best" scaled values for specific genomes within a reference database and distribute those sketches in the initial .zip.

ref #1524 (comment) - we return original sketches from search, not downsampled ones - so this gives us options to do the initial search at higher scaled and then go lower as needed.