Implement new virus half-life formula (humidity & temp. dependent)

Some changes to the model to implement the formula for Covid-19 half-life from Dabisch et al..

Test results have been adjusted to match the outputs with the new formula. In most test cases the difference is very minor (a few %). @lrdossan and I have checked the new test values, but this may be worth an extra set of eyes checking before we merge.

This change allows for better risk modelling when connected to a real time sensor input that measures temperature and humidity.

NB a hard limit is included on the maximum half-life of 6.43. This is consistent with our previous implementation for most cases, since our humidity model is static. Without the limitation the formula can give a negative decay rates in some cases, which after discussion with the author we will ignore (replaced with ln(2)/6.43).

The results of the new humidity model:

the REST API is also updated by exposing inside_temp parameter