Flow over heated plate uses Aitken acceleration (inconsistent)

[MakisH]

See https://github.com/precice/tutorials/blob/0482e5b4d3e1604dd031d2b113a2aaa63733c9a8/flow-over-heated-plate/precice-config.xml

In contrast, the -nearest-projection variant uses IQN (which we should be showcasing): https://github.com/precice/tutorials/blob/0482e5b4d3e1604dd031d2b113a2aaa63733c9a8/flow-over-heated-plate-nearest-projection/precice-config.xml

I don't know why we have this difference (I would need to do some digging into relics), but using the same acceleration config as the -nearest-projection variant seems to be working.

[davidscn]

IIRC the Aitken acceleration was way faster than the IQN acceleration, as the solver iterations are very cheap. I don't get why we apply everywhere IQN acceleration. In my opinion, the tutorials should cover all available configurations as good as possible.

[MakisH]

We don't have to use IQN everywhere. In fact, some variety would also be useful for the system tests.

However, we should document the motivation for anything that looks inconsistent. We could even see this as a problem with the flow-over-heat-plate-* cases that use IQN (and change those to Aitken).