Implement scaling by temperature ratio

[evertrol]

For one (or a few) scenarios, the model of interest (EC-EARTH) can't match the overall CMIP global annual temperature change: it is limited to a maximum year year of 2100, which may be below a warm scenario for 2085.

In this case, a factor is deduced, the ratio between the required temperature increase (global CMIP delta-T in, say, 2085), and the actual temperature that the model can reach. This factor later has to be applied (linearly) to all estimated values for e.g. (seasonal) temperature increase, precipitation change etc.
This does assume that the factor is small enough that all derived values change linearly with the temperature (that is, it's a first order effect only).