Update to analysis_common_envelope

[themikelau]

It is time to update this analysis module.

Changes to analysis_common_envelope:

• Option to output extra quantities to ".divv" files replaced with a proper implementation, where an arbitrary number of extra quantities can be written to ".extras" files, which can be read by splash (see https://github.com/danieljprice/splash/pull/87)
• Most important options are made compatible with simulations run with radiation.

Changes to other parts of Phantom:

• Functionentropy in eos module: Add option to specify radiation temperature directly, which is useful if using radiation and Trad is stored.
• radiation_utils: Redefine radE_from_Trad (output is energy per unit volume) to radxi_from_Trad (output is energy per unit mass), the latter being more useful. Likewise, the inverse function is also changed from Trad_from_radE to Trad_from_rhoxi. The rest of the code has been adjusted to be consistent with these new definitions.
• eos_idealplusrad: Add functions to calculate gas energy and radiation energy separately.

[themikelau]

The test suite caught a bug with the cons2prim test. There was an issue because I have lumped some changes to entropy calculation in this pull request. Namely, I have ensured that any temperature solving uses the gas constant, Rg, instead of kb_on_mh. Mixing the two constants causes a discrepancy great enough to fail the cons2prim tests. But I forgot to change this in the temperature solving in get_p_from_rho_s. This has been fixed now. Can @Fitz-Hu check if these changes are correct?

[danieljprice]

See comment on splash PR, would be better to use .cols file which is an existing splash feature

[danieljprice]

Also the cons2prim stuff should be pretty well unit tested so happy to approve that