Add lower-resolution mesh for BFS

[jcallaham]

Added a new mesh that has about 1/3 of the elements of the full-resolution mesh but is pretty close to matching validated results. Here is a time series of the fluctuation KE (scaled by domain size) with the same random seed in the new mesh and "fine":

The new mesh has outflow length L=15 compared to L=25 for the "fine" mesh. We can't really compress it much more than that without losing significant resolution, since the upper flow doesn't reattach until x=10 or so. The new mesh has about 81k elements, compared to 223k elements for the fine mesh.

Coarse mesh:

Lower reattachment point: 5.66 (L0)
Upper separation point: 4.46 (L0)
Upper reattachment point: 10.09 (L0)

Fine mesh:

Lower reattachment point: 5.84 (L0)
Upper separation point: 4.63 (L0)
Upper reattachment point: 10.19 (L0)

Boujo & Gallaire (rescaled to our length scale):

Lower reattachment point: 5.9 (L0)
Upper separation point: 4.7 (L0)
Upper reattachment point: 10.3 (L0)

The examples/step/solve-steady.py script also includes stagnation point calculation.

[ludgerpaehler]

Looks good to me