Forcing and drag for the ShallowWaterModel

[milankl]

fixes #151

1. Adds a forcing term of the form

$$\mathbf{F} = (F_u(\theta),0)$$

which only depends on latitude $\theta$ and mimics a jet stream in mid-latitudes that sits above the shallow water layer.

(6h after starting from rest, the perturbations are from orography)

2. Adds a quadratic drag term of the form

$$\mathbf{D} = -\tfrac{c_D}{H}\vert \mathbf{u} \vert \mathbf{u}$$

with a drag coefficient $c_D$ and the shallow water layer thickness $H$.

[milankl]

After 5 days

After 10 days

After 50 days

And vorticity after 50 days

[milankl]

With $c_D = 10^{-5}$ the wind speeds are with 150m/s regularly somewhat too strong (delibrately testing at low T31 resolution)

[milankl]

For $c_D = 2\cdot 10^{-5}$ this decreases to about 125m/s

[milankl]

Or similar for a longer forcing time scale (=weaker) forcing, which also makes the northern jet weaker and less stable

[milankl]

Last commit makes the forcing weaker by 50% looks good (T85)

[milankl]

@ningliu-iga with this PR we can use the ShallowWaterModel for very long integrations, the forcing provides an energy input into the system, the drag removes it. Running it for O(years) seems like we're getting a stable climatology with turblence plus a bunch of Rossy waves that travel along the jet streams on both hemispheres. Shown is day 100 to day 200.
We can probably adjust how turbulent vs how wavy it's supposed to be by adjusting forcing/drag/layer thickness somewhat.

out.mp4

@maximilian-gelbrecht I think we talked about a climatology in the ShallowWaterModel before, this PR addresses this.

[milankl]

Run with

spectral_grid = SpectralGrid(trunc=85,nlev=1)
quadratic_drag = SpeedyWeather.QuadraticDrag(spectral_grid,c_D=1e-5)
jet_stream_forcing = SpeedyWeather.JetStreamForcing(spectral_grid)
model = ShallowWaterModel(;spectral_grid,drag=quadratic_drag,forcing=jet_stream_forcing,initial_conditions=StartFromRest())
simulation = initialize!(model)
run!(simulation,n_days=100,output=true)