WR spectra rescale by luminosity

[ramseykarim]

In the catalog module, the WR spectra and mass loss rates need to be scaled with the observed (or assumed) luminosity; the values from the model are luminosity-invariant. From their website:

Each grid has a two-dimensional parameter space, namely the stellar temperature T and the "transformed radius" Rt. The latter is a function of the mass-loss rate and the stellar radius. Thanks to an approximate scaling invariance of WR atmospheres, the same model spectrum can be applied to stars with different luminosities, but same T and Rt. When doing so, the absolute fluxes must be scaled proportional to L, the radius R proportional to L^1/2, and the mass-loss rate proportional to L^3/4.

This needs to be addressed and tested. Testing should check that:

1. The WR spectra produced by the current version of the code do not integrate to the observed luminosity, but instead to logL = 5.3 (the PoWR default)
2. The rescaled spectra don't upset any published results
3. The mass loss rates I used were from the observations/modeling rather than from PoWR, and that the rescaled PoWR mass loss rates are not in huge disagreement with the observed/modeled ones

This can probably be fixed within powr.py since it is specific to the PoWR grid; the other parts of the code should not be aware of this peculiarity.

[ramseykarim]

Issue solved. The fix ended up in stresolver.py rather than powr.py because, while the peculiarity is with PoWR, powr.py doesn't know the real luminosities and has no reason to handle them. Only stresolver.py should know about the real (sampled) luminosity. The new FUV and ionizing fluxes for WN6ha seem reasonable, they jumped by log10~0.7, which makes sense since the normalization jumped from log10 = 5.3 to 6.06 (1.15e6, the value from Rauw)