Deal with missing OTT ids better

[jar398]

For induced_subtree and mrca methods, we yield a 400, where in the v2 API we returned the list of missing ids. In spite of pre-release review, this has turned out to be an inconvenience for users. We need to either restore the not-an-error-on-missing-id behavior, or introduce a validation service that people can call beforehand as a filter.

[josephwb]

See also: ropensci/rotl#31

[josephwb]

It sounds like the new propinquity results will include "broken" taxa. If that gets ingested into treemachine then it will again be possible to distinguish invalid (i.e. pruned from taxonomy or straight up non-existent) vs. valid-but-not-in-tree ids. Right?

[josephwb]

Note that this also is relevant in the curator when checking the mrca of the ingroup against the synthetic tree: if any of the mapped tips are excluded from synthesis (extinct taxa, in my case), an error is returned.