Refinements to the committee approach

[simonwengert]

Some things to be discussed / not yet fully implemented:

• Read in calibrated parameters from disk (avoid recalibrating) + writing calibrated parameters to disk
• Generalize scaling as sigma_scaled = alpha * sigma**(gamma/2 + 1) (with default gamma=0)
• Add tests