The problem is that I have no way of sorting the remaining unmemorized vocabulary for priority without altering the current file (which is broken into lessons and reviewed periodically )beyond what I would prefer.
For the time being I can use your suggestion above to artificially move the unmemorized but un-useful words up to level 5 and where they can be more easily ignored.
You are right, there is no concept in Stackz at the moment of classifying the entries according to "relevance" or "importance". Every entry is considered as of equal relevance.
Your idea certainly will work fine, but only with the ColorMode "RepeatAdvisor", with difficulty estimation set to "promotion level". In this setting, the repetition interval depends on the card's refresh date and the promotion level (longer for high promotion levels, see Repeat Advisor ColorMode detail settings). The other ColorModes are based on the card refresh date and/or past success/failure counts, which does not depend on the promotion level.
Background: The ColorMode visualizes the card's "good or bad" state with a color, and the study tools process the selected cards in the "bad to good" order. The ColorMode "RefreshDate" simply uses the last positive test for classification (the oldest card is processed first). The RepeatAdvisor extends the RefreshDate algorithm by highlighting (and processing first) the cards that are "overdue", i.e. "current difficulty estimation is worse than the threshold (for this level)". This corresponds to the well known "spaced repetition" algorithm.
The question is just whether you want to have the "not so important" words in your list in the first place. You could also move those cards to different lessons (e.g. "Lesson 1 optional", or "all optional words") or even to a special file. But then you would have to decide yourself when you want to
study those words, as opposoed to the algorithm telling you to
look at a less relevant word (while you are
studying relevant words).
Future considerations in the Stackz system go in the direction of adding "flags" for the entries. A parameter "relevance" could be an extended flag (or parameter). Such developments have low priority at the moment, and we also try to avoid overloading the system.