speed up chabrier1998 screening by adding `` entries

[zingale]

Currently, in chabrier1998, we do:

  Real Gamma1 = Gamma_e * std::pow(scn_fac.z1, 5.0_rt/3.0_rt);                                                                                          
  Real Gamma2 = Gamma_e * std::pow(scn_fac.z2, 5.0_rt/3.0_rt);                                                                                          
  Real Gamma12 = Gamma_e * std::pow(zcomp, 5.0_rt/3.0_rt);   

we should move these pow's into screen_factors_t, perhaps only enabled if we are using this screening routine. This will save on a lot of expensive pows

[yut23]

The quantum correction terms also use a lot of pows, which we should probably switch to nested evaluation like in screen5.

[zhichen3]

I'm actually debating on whether those terms are needed. These quantum correction terms were not in the original chabrier&potekhin paper. They were introduced in Calder's paper, so I added them. Removing them might give better convergence with NSE states too, although I never tested it.

[zingale]

I think that this has been addressed now