Scam-V: Reorganise modules to couple observation model and side-channel measurement

[pablobuiras]

Once an attacker/observation model is fixed, this should also determine the code we use to measure the channel on the hardware. So we would like to reflect this in the module structure somehow. The driver could be parametric on a 'side-channel' module, which would define both the add_observation function and a particular instantiation of embexp-driver (or the part that controls the measurement).
NB. Make sure this is general enough to support observation model refinement.