Small dyn spn adaptations

[pat-schmitt]

Here I am suggesting some changes in the dynamic calibration functions:

• I changed the default min ice thickness from 10 to 2. I found this in a recent study that with the old default of 10 you are always on the higher end of the target area error boundary. This makes a larger difference for smaller (thinner) glaciers, on the large scale I do not expect to see large differences. Open for discussion! (if we change this should be added to whats-new)
• I slightly changed the order in the code to avoid wrong diagnostic messages in gdir.get_diagnostics() if something fails. But all functionalities stay the same.
• I renamed t_bias to t_spinup in the dynamic calibration procedure, to avoid confusion with the temperature bias of the mass balance model (suggestion from @rodaguayo)