Support shearing the tree to feature metadata in tree-plot

[fedarko]

Currently, --p-shear-tree is only available for the qiime empress community-plot command (where it shears the tree to just the tips contained as features in the table).

@antgonza brought up that it would be nice to support shearing the tree to the feature metadata, for example when qiime empress tree-plot is passed a massive tree and slightly less massive feature metadata information.

Things we will need to consider for this issue:

• Should this option also be available for community-plot? If so, how will this interact with the existing shear-to-table option there? (We might need to rename that parameter to something more specific than shear-tree, at least, just to avoid confusing users.)

  • If both shearing options are available for community-plot, we'll definitely want to use an explicit "order" for how shearing is done. There are probably going to be a few annoying cases where applying one shearing method causes the other shearing method to fail, or something, so testing this will be important. (For example: if the feature metadata is only defined for tips not present in the table, but both shearing methods are provided, then things should explode because no tips would remain. Or, if the only features shared by the feature metadata and table are all zeroes in the table, then this should still fail, because empty features / samples are filtered out before creating the visualization.)

• Feature metadata doesn't have to refer to tips -- it is possible to pass Empress a feature metadata file with only internal node metadata. I imagine that, in this case, shearing would only be done on the tips within the feature metadata, and if there are no tips in the feature metadata then an error should be raised.

• probably other stuff i'm forgetting

After writing this out, this seems like it might be kind of complex to implement alongside the existing shearing method... so maybe just doing this for tree-plot at first would be reasonable.

[fedarko]

Closed by #428.