Add a Forward/Backward time stepping test

[xylar]

There were issues with the FB time stepper in E3SM-Project/Omega#428 that were only revealed by CTests + NaN detection, not regression tests because none of the Polaris tests in the omega_pr suite use FB. We should remedy this and test all time steppers.

[xylar]

The issue in E3SM-Project/Omega#428 turns out not to be related to FB but it still wouldn't hurt to have a test for that stepper.

[cbegeman]

@xylar Tests for FB and RK2 are on the integration flow chart/gantt charts. If you have time before I can get to this, you can assign yourself. As we discussed in an Omega meeting, the intention is that these will be barotropic tests because we do not anticipate using these schemes outside the barotropic part of the SE scheme.

[xylar]

That's fine. I don't think this is pressing. It initially seemed like the FB scheme was seeing bugs that RK4 wasn't but the real issue was #639. So I don't think this is pressing.