If there is a unique forecast, extreme limits and tercile limits must be vectors. Now 1-D array is cot accepted. Observations are proven to be a vector or a numeric value.

Corrected bug in labels of tercile/extreme probabilities. Now uses first panel positions for all labels.

Removing aperm manipulations before the Apply function in CST_Calibration.R and partly reversing corrections of Verónica.

Suggestions of Verònica implemented. More specifically: 1) implementation of multiApply as wrapper, 2) removal of ... 3) warning when multi.model is used, 4) adaptation of the manual by addition of bias method, discussion of use of multi.model boolean, change of adequate reference to 2011 paper for evmos method.