only feature ids present in the table should be clustered

[gregcaporaso]

I helped a collaborator run cluster-features after importing sequences and a biom table that were processed with a non-QIIME 2 pipeline, and we had the surprising result that the resulting FeatureData[Sequence] had more records in it than there were features in the FeatureTable. It turned out that the input FeatureData[Sequence] had sequences for features that had been filtered out of the input FeatureTable. We should probably raise an error if the set of feature ids isn't the same between the two inputs.

This is a bit of an edge case, so isn't really high priority (this will probably primarily be applied to FeatureTable and FeatureData[Sequence] artifacts that are generated through a QIIME 2 pipeline).

[gregcaporaso]

Addressed in #11.