Fix hold iterations

[ftsamis]

Fix the following issues:

• There were cases where the hold_iterations mode was triggered by the convergence_strategy criteria, and the simulation would get in an infinite loop.
• Improve counting the simulation iterations in the log.
• Also count the last iteration in the final summary message.

[ftsamis]

@wkerzendorf If the hold_iterations in the configuration are set to 0, and an iteration converges, should the simulation be over?

[wkerzendorf]

@ftsamis I would say hold-iterations >=1 as a constraint

[ftsamis]

@wkerzendorf Done

[yeganer]

I don't know if there are plans for further changes to the Simulation logic but while we're at it, I suggest to make all the variables dealing with convergence and iterations local.
Or, if Simulation.iterations_executed are really needed, we should initialize them in init and not there.

[unoebauer]

Should be merged before v2.0

[ftsamis]

#652 has a better version of the fix, along with all the other changes I introduced during the summer. Let's close this and focus on #652.