Feature/particle integration test

Adds an in-depth convergence test for particle tracking of a stationary field.

In the below figure "o" stands for "solver order", "n" for interpolation neighbours and "m" for interpolation smoothness.

Notes:

• I also added a classic Runge Kutta solver for particles, to emphasize effects of interpolation. The plot suggests that the implementation is correct (I need to see order 4 convergence to believe the code is correct).
• The number of particles is relatively small now because otherwise the test is too slow in DEBUG mode. There seems to be a separate issue with optimization that I'll need to address.
• The effects of the interpolation are only visible if the particle trajectories are long enough --- i.e. particles need to traverse a few grid cells in order for cell-crossing effects to show up. This is why there are so many iterations...

assigned to @toba

changed the description

added 1 commit

• 97e78f50 - cleans up error plot

Compare with previous version

Latest commit cleans up the error plot.

approved this merge request

mentioned in commit 9c592f2b

merged