Changing R version to 3.4.2

See merge request !45

Handling of case of insufficient data in CST_Calibration

See merge request !44

Dev plot forecast independent terciles, for January 2020 release

See merge request !47

CST_BEI_Weighting working with NAs

See merge request !46

CST_MergeDims function to merge two dimensions from an array

See merge request !43

Merge branch 'dev-PlotForecast-independent-terciles' of https://earth.bsc.es/gitlab/external/cstools into dev-PlotForecast-independent-terciles

1) Introduction of ncores as extra parameter. 2) Introduction of Calibration as a new function (replacement of .calibration.wrap). 3) removal of var.fc

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.