Diff abund: better support for longitudinal, metabolic data

[d-callan]

Maaslin is currently configured to use default params, but we know from the original publication that that is sometimes not necessarily the most appropriate thing to do. We should consider how to automagically use different params for different data types.

related, we could consider offering some level of user control as well?