Calculate computation times for each CRAN version

In order to keep trace of the evolution of the computing times after modifications of the packages, I think we should automatically calculate the computation times in a similar way as done in Coron et al. (2017) (https://doi.org/10.1016/j.envsoft.2017.05.002), Table B.3.

I propose that:

• each former CRAN version is tested now

• each future new version of the package is tested (before submission for detecting problems(?) and after publication on the CRAN)

• all tests done in Tab. B.3 are performed (i.e. runs and calibrations over 10 years for each hydrological + snow model combinations, 100 tries)

• other functions like Imax, PE_Oudin, DataAltiExtrapolation_Valery are tested (configuration to determine)

• at least one Windows machine and one Linux machine are tested, Mac if possible.