Better handling for temperature and pressure of a Simulation

Currently, as seen in #1270, if the temperature or pressure of a simulation is not set, this is not caught until the simulation falls over somewhere (e.g. when minimising).

From #1273:
> But you are definitely right that it makes no sense to use MDMC without setting the temperature sooner or later. And I do not see an advantage to running the minimisation without setting the temperature.
>
> Having said that, is there an easy way to change the temperature? Because I can imagine the user running minimise at 300K, then at 100K, and then running the simulation itself at 135K. So, we should definitely allow this approach.

The Simulation should be changed so that it cannot be created without setting temperature or pressure (unless someone can give a good use case for this). In addition, we should check that the temperature can be accessed and changed if desired by the user.

Connected issues : #1250, #786