[cli] add --resume mechanism

[phisad]

It would be nice when aborted partial runs can be resumed, that is, if:

• the results folder is already existing, and in there
• the model folder is already existing

Then the framework should identify

• which instances have already been run, and
• remove them from the schedule/skip them during the resumed run

[phisad]

Still relevant.

Naive approach:

• parse result folder

Maybe a better approach:

• store run information in run.json