VR if the species never change

[laurenmh]

The variance_ratio calculation for this is NA, which is a problem for the lower and upper confidence - means that it won't run at all if a replicate has this problem.

Options to fix it:
fix with quantile (na.rm=T) - not ideal because it masks the problem.
highlight that those plots should be removed.

[laurenmh]

added check_sppvar to utilities and to the variance_ratio_matrixdata function, stops analysis and explains why