Active lower layer in reduced gravity mode

[edoddridge]

Add the capability to have a quiescent upper layer and flow in the bottom layer - an "upside down" 1.5 layer model.

[codecov[bot]]

Codecov Report

Merging #241 (e99c517) into master (a837cf5) will decrease coverage by 0.68%. The diff coverage is 47.05%.

@@            Coverage Diff             @@
##           master     #241      +/-   ##
==========================================
- Coverage   75.10%   74.41%   -0.69%     
==========================================
  Files          20       20              
  Lines        1390     1419      +29     
  Branches       81       85       +4     
==========================================
+ Hits         1044     1056      +12     
- Misses        299      313      +14     
- Partials       47       50       +3     

Impacted Files	Coverage Δ
src/aronnax.f90	30.70% <0.00%> (-2.01%)	:arrow_down:
src/model_main.f90	84.07% <ø> (ø)
src/bernoulli.f90	41.17% <37.50%> (-0.93%)	:arrow_down:
src/state_deriv.f90	80.00% <75.00%> (-1.82%)	:arrow_down:
aronnax/driver.py	92.43% <100.00%> (+0.12%)	:arrow_up:
src/time_stepping.f90	58.94% <100.00%> (+0.88%)	:arrow_up:

Help us with your feedback. Take ten seconds to tell us how you rate us. Have a feature suggestion? Share it here.

[edoddridge]

Tests added:

• an active lower layer simulation in which the layer thickness perfectly mirrors the bathymetry. This causes the bounding isopycnal to be flat, and therefore there are no pressure gradients and no motion.
• an active lower layer simulation with sloping bathymetry and an initial condition with uniform layer thickness. As expected this leads to a gravity driven rearrangement and the fluid moves down slope an the bounding isopycnal flattens over time.

The coverage stats aren't correct. For some reason the execution counts aren't accumulating across all of the tests. I will open an issue about this. In the meantime, this PR can be merged.