Get Omega reference density from polaris.constants when computing z-tilde

[xylar]

With this merge, we no longer support a polaris config option for rho0, instead relying on seawater_density_reference that will soon come from the Physical Constants Dictionary (PCD).

This is the Polaris equivalent of E3SM-Project/Omega#358.

Checklist

• Testing comment in the PR documents testing used to verify the changes

[xylar]

@alicebarthel, could you have a quick look when you can? No existing Polaris tests use the functions from which I have removed rho0 as a parameter, but #465 will. I'll rebase that branch onto this one.

[xylar]

Testing

Using tests from #465, I'm seeing all passing on Chrysalis with Intel:

0:01:40 PASS ocean/horiz_press_grad/salinity_gradient
0:01:32 PASS ocean/horiz_press_grad/temperature_gradient
0:01:32 PASS ocean/horiz_press_grad/ztilde_gradient
Total runtime: 0:04:45

It doesn't look like the switch from rho0 = 1035 to 1026 made any difference in terms of number of iterations needed to compute the pseudo-thickness initial condition either.

The omega_pr suite also passed on Chrysalis with Intel (even without #488!).

[alicebarthel]

@xylar This looks good to me based on visual inspection of the code changes.
Did you need any extra tests on this?

[xylar]

@alicebarthel, nope, just wanted a quick visual inspection like you did. Thanks!

[xylar]

We missed

polaris/polaris/ocean/conservation.py

Line 4 in 9f11d07

rho_sw = 1026.0

Should follow up sometime soon...

[alicebarthel]

Good catch, @xylar!