Add protection against negative pressure in HybridEos

[ncorsobh]

Proposed changes

Floating point issues can lead to negative pressures in HybridEos. This quick PR puts a safeguard against this.

Edit: delving into fixing the unit-tests has revealed another related issue, so I've split the PR into two commits. For the first commit, the specific_internal_energy_lower_bound and specific_internal_energy_upper_bound functions in 1D EoSs are supposed to provide these bounds as a function of rest mass density. Since in the case of a 1D EoS the internal energy is not a free parameter when density is known, these bounds can only be the exact specific internal energy returned by the 1D EoS.
The second commit deals with the original floating point problem with HybridEos.

Upgrade instructions

Code review checklist

• The code is documented and the documentation renders correctly. Run
  make doc to generate the documentation locally into BUILD_DIR/docs/html.
  Then open index.html.
• The code follows the stylistic and code quality guidelines listed in the
  code review guide.
• The PR lists upgrade instructions and is labeled bugfix or
  new feature if appropriate.

Further comments

[nilsdeppe]

Uh oh, tests are unhappy :(

[ncorsobh]

@nilsdeppe updated PR, everything should be fixed now. Let me know if the split into two commits was unorthodox and whether a separate PR would be more appropriate.
Also @isaaclegred maybe you could take a look too since this is digging up some EoS changes?

[nilsdeppe]

I think this might break @jyoo1042 's simulations with a 1d EOS. I believe for primitive recovery we need the thermal part effectively unrestricted and trying to restrict it caused the simulations to fail. Isn't what you would like for the 2d EOS to have a density-dependent bound, not the 1d EOS? For a 1d EOS the thermal part doesn't mattr

[ncorsobh]

@nilsdeppe PR updated to reflect our last discussion. For posterity's sake, we had agreed that the 1D specific internal energy bounds should lose their density dependence entirely, and that HybridEos should set its lower bound for specific internal energy using cold_eos_.specific_internal_energy_from_density.
I overwrote the first commit rather than pushing a fixup commit since it was pretty much completely overhauled anyway.
For PolytropicFluid and PiecewisePolytropicFluid, the special upper bounds were chosen to be consistent with the upper bounds set on rest mass density.
@jyoo1042, if the PR is merged in its current state, the main (likely only) thing you'll need to change is to remove density from the function call specific_internal_energy_lower_bound (and upper bound) if you're using a 1D EoS.