I would like to ask if SEM offered by MPlus applies variance based PLS according to Wold (1989) or if it is covariance based? As far as I know, PLS makes sense if sample size is small and data is non-normal with much multicollinearity.
It is covariance based in line with traditional SEM. PLS is something quite different, estimating different parameters. SEM is factor-analysis based, while PLS is principal-component based. If your model is SEM, I would not turn to PLS just because of small sample size, non-normality, or multicollinearity