LOO error when k_threshold = 0.7 fits model that drops sole observation of factor level

I am getting the same error (regarding new factor levels) when using kfold with leave one group out. The model looks like:
y~factor+x+(x|group). I use ‘group’ to generate my folds but recieve an error when ‘factor’ has a new level. I can try to post a reproducible example but seems from the discussion that the developers are aware and will be addressing this issue. Has it been addressed in the dev. version?

I am using
R-3.4.3
rstanarm 2.17.4
loo 2.0.0