Fixed subcycling in implicit coupling (for regular simulations)

[boris-martin]

(Depends on #105 )

Fixes https://github.com/precice/calculix-adapter/issues/9. Essentially, checkpointing stores more data than previously (not only positions, but velocities and a few more things), because those additional fields must be stored when going back from more than an increment (time step). So previous implementation worked only when there was one step per window.

Still WIP:

• [x] Refactoring & cleaning
• [ ] Outputting once per time step instead of once per window (using the buffer from #105)
• [ ] Allow user-defined output frequency (once every N steps)
• [ ] Testing on more cases, including thermal cases (only perpendicular flap so far)
• [x] Check for redundancy (I may have copied more data than necessary?)