Remove junctions with Ns from cloning

[Jason Vander Heiden]

Junction sequence with multiple N characters can completely screw up a clone due to single linkage. We should remove junction sequences with N character somehow. Some options:

1. Failing such sequences.
2. Clustering without them and then assigning them after the fact to a cluster they match. Randomly perhaps, as they are likely to be zero-distance from multiple different sequences. Might be able to resolve by assigning to largest clone.

[Jason Vander Heiden]

Added --maxmiss argument in eccbc5f to control failure criteria for junctions with missing data.