New options

New parameter L0 which corresponds to the maximum number of variables to be selected on each of the R components. This is for objects mddsPLS and perf_mddPLS, under name L0s which is a vector of integer values corresponding to the to be tested vvalues of L0in the cross-validation procedures.
Nice visualisations for components and super-components through barplot or heatmap classical R functions. Extra variance explained is accessible also, thanks to classical OLL or RV paradigms. Those are accessible for mddsPLS objects under plot method.
Methods summary for objects mddsPLS and perf_mddPLS describing different components and missing path for mddsPLS objects and cross-validation results (time, convergence,...) from perf_mddPLS objects.
New parameter var_selected from mddsPLS objects permitting to know the selected variables and their coefficients on the components and super-components.
New parameter keep_imp_mod from mddsPLS objects permitting to know whether the imputation models must be kept and returned to the user. It makes objects heavier but can be useful for interpretations.
New parameter NZV from mddsPLS objects permitting to fix the threshold above which coefficients are considered as null. Might not be changed by users.
New parameter getVariances from mddsPLS objects permitting to know whether variances must be computed, quite time consuming but important for interpretation.

Code enhancements

New initializations of missing values: marginal block missing value imputations.
Use of Rcpp classical functions decrease time of computations.
Modification of the Tribe Stage, use the all Y and not its projected component S for estimating missing values.
Use of biased variance estimators as to fit with the Python version.