provide more comprehensive/useful database benchmarks

[ctb]

In #2013, I pull numbers from a smattering of numbers from previous benchmarking efforts - specifically,

#1958
dib-lab/2020-paper-sourmash-gather#47
#1530

that I summarized like so:

Search type	Query	Database	Max RAM	Time
gather	Bacterial genome	GTDB complete (280k)	1 GB	6 minutes
gather	Simple metagenome	GTDB reps .zip (65k)	2 GB	6 minutes
gather	Real metagenome	All Genbank (1.2m)	100 GB	3 hours
lca summarize	Simple metagenome	GTDB reps .sql (65k)	400 MB	20 seconds
lca summarize	Simple metagenome	GTDB reps .json (65k)	6.2 GB	1m 20 seconds

while these are all well and good, there are a few problems:

• we don't supply the details of the queries
• we don't provide numbers for search and prefetch
• we don't really talk about the various other operations at all (e.g. manifest loading, and so on)

it would be nice to update with a better discussion!