Option to exclude seeds or clusters with low abundance

[torognes]

A user have suggested an option to exclude seeds sequences or clusters with low abundance (e.g. singletons). This could be applied to the dereplicated input sequences or to the output clusters.

Even if this does not save much time, a large amount of output could be avoided, especially when there are lots of singletons.

[torognes]

To be a bit more precise: clusters with centroids/seeds that have a low abundance could optionally be excluded, but low-abundance sequences could be part of other clusters. This cannot be done by simply filtering input or output based on abundance.