Simplify convergence strategies

[unoebauer]

Currently, there is very little difference between the two supported convergence strategies, damped and specific. Also, as noted in issue #695, a number of properties of the convergence_strategy are unused. This PR tries to simply the convergence strategy handling by

• merging the two existing strategies into one, which will be called damped (the name describes the current strategy to update the plasma variables best)
• introduce support for currently unused properties, in particular for lock_t_inner_cycles, t_inner_update_exponent
• remove remaining unused properties
• introduce new property that controls whether the simulation stops before the specified number of iterations if the plasma quantities converged or not

[unoebauer]

@wkerzendorf - what do you think? I still have to check the different tests and probably need to update some reference data...

[wkerzendorf]

@unoebauer In general, I agree - it means that we will break backward compatibility (which we have done in the past and which I'm in favor off).

[unoebauer]

True - but one only has to change the type from specific to damped. All other options are still valid under the updated convergence strategy handling

[unoebauer]

Ready to be merged from my point of view.