consider adding --max-containment option for some sourmash utilities

[bluegenes]

Given the work we've been doing with maximum containment --> genome distance metrics, there are some places where it might be useful to return the maximum containment between signatures, rather than directional containment. This could be enabled with a --max-containment option for certain commands.

Cases I can think of:

• compare
• search -- e.g. my current desired use case is sourmash search --max-containment --best-only to find the best match for an input sig (or list of input sigs) to a list of cluster founders.

--max-containment is likely not useful for gather-style metagenome applications, primarily bc the direction of containment (intersection/reference genome hashes) is already ideal.

[ctb]

it may not be possible to do this efficiently on SBTs - thinking out loud,

• SBTs provide Jaccard similarity and containment searches for a query Q in a database D
• can there be a match S in D that has low Jaccard similarity and containment but high max containment? this would be a match that cannot be found using current SBT, but would be reported for high max containment.
• I can't think about that clearly, but maybe a way to rephrase it is to ask about doing containment searches (Q in D) and ((all d in D0 against Q) and taking the best match?

perhaps this issue on reverse containment is related? #1198

[ctb]

so uh that was silly of me,

yes, this is straightforward. the SBT guarantees results by largest number of hashes shared; max containment would just be changing the denominator.

[ctb]

so, uh, not quite as straightforward as I'd hoped, but we'll see :)