Reorganise "hydrostatic" vertical slice tests

[tommbendall]

I had always been confused about the "hydrostatic" vertical slice tests, thinking that they were only for use with hydrostatic equation sets.

But reading the Melvin et al 2010 Cartesian ENDGame paper, it's obvious that the test cases can be used with either hydrostatic or non-hydrostatic equation sets. The hydrostatic tests are larger scaler, and so the solutions should be similar between hydrostatic/non-hydrostatic equation sets, whereas the non-hydrostatic tests have a smaller spatial scale and so should show differences.

This PR therefore does some reorganisation and several small related tasks:

1. the Skamarock-Klemp gravity wave tests and the mountain tests are now implemented with a hydrostatic argument, which would turn on the "hydrostatic switch". This means we can run all of these test cases with hydrostatic/non-hydrostatic equation sets
2. I have updated the final time for the non-hydrostatic gravity wave (I don't know why this was 3600 s, when in the literature it is always 3000 s -- @atb1995 pointed this out to me)
3. I have tweaked the plotting script for the non-hydrostatic gravity wave so that the wave does not wrap around the periodic boundary
4. I have moved the mountain test to the case studies repo, in the sister ticket Hydrostatic/Non-hydrostatic Tests gusto_case_studies#34
5. I have added the Schaer mountain test case so that there is still a mountain test in this repo
6. I have implemented a spinup_steps option for the semi-implicit quasi-newton time stepper, which sets alpha to 1 for a specified number of steps at the beginning of the model, which should help with the initial difficulties over orography