Generalise stellar wrapper, dummy star module

[nichollsh]

This PR adds a dummy star module. It calculates the stellar spectrum and bolometric luminosity assuming that the star emits like a blackbody with a temperature and radius specified by the user. Added dummy_star.toml configuration to demonstrate this, but there isn't a test for this yet. Closes #202.

It also reworks the stellar module handling to generalise calculations between the mors and dummy modules. This mostly means moving functions into src/proteus/star/ and calling these from proteus.py. This also meant moving a lot of variables in proteus.py into the Proteus class, which will also be useful for later work (e.g. #204, #162).

Minor things:

• Fixes numerical issue when volatile elemental inventory is small. Sometimes floating point errors mean that 0 kg of a given element can become ~1e-13 kg or so, which is problematic.
• Renamed Lbol config variable to lum_now, to make it clear that this is the modern luminosity of the star.
• Added phys.py for containing physics functions.
• Removed start_proteus.py since it is made redundant by the CLI.
• Fixes some of the deprecation warnings when accessing the config.

[nichollsh]

That makes sense - I guess this is covered by #167 ?

[nichollsh]

Thanks both!

[lsoucasse]

That makes sense - I guess this is covered by #167 ?

Yes!