Input sequence names identical to sequence name in Rfam alignment will cause R2DT to fail

[nawrockie]

With the 'Rfam' CM library, R2DT uses the cmalign --mapstr and --mapali <s> options where <s> is the training alignment file used to build the CM being used. This allows R2DT to propogate any pseudoknots from the training alignment to the output diagram. However, if an input sequence has a name that is identical to one of the sequence names in the training alignment, then cmalign with rename all the sequences (to avoid duplicates) by prepending a numerical index. This causes R2DT to fail because it relies on being able to identify and remove the original training sequences from the output alignment, and if their names have changed it cannot do that.