Handle no-persistent-ID-for-given-distance case

[aofarrel]

If not starting fresh, and there are a nonzero number of persistent IDs for (say) 20 and 10, but no samples whatsoever have persistent IDs at 5, the process clusters script will crash pretty quickly because no "rosetta stone" file will be created for that distance.

This is not the same as these cases, which are handled just fine:

• A sample had a persistent ID at 20 and 10, but not at 5, and now clusters at 5 (ergo needs a new persistent ID at 5)
• The ad-hoc "starting fresh" case, where nothing has a persistent ID

[aofarrel]

Currently the crash is caused by a failsafe for nondeterministic behavior in the perl script. That failsafe should probably be either removed or toggled, but even then we would still crash not long after.