zk/cocos

[ZanderKeith]

Our goal is to deploy on several different devices, so we need to carefully handle all the different coordinate conventions used. I propose converting everything to COCOS 1 and use that internally. This PR does that like so:

1. Two helper functions detect_cocos and convert_cocos.
   Given an arbitrary eqdsk, detect_cocos will determine which COCOS is being used.
   Then, convert_cocos will use the proper sign mapping to convert the eqdsk into the target COCOS

2. On creation of EquilibriumField, the eqdsk is automatically converted to COCOS 1

Now during field line tracing we can always be confident we know the correct vector for the field lines based on the signs of B0 and Ip.

One lingering thing which we might want to address is our definition of 'helicity'. At the moment, we're saying helicity is synonymous with the sign of m/n, where positive values are parallel to the field lines and negative values are antiparallel. I'm not sure this is technically correct but we can simply rename this to be something else (like mn_sign or whatever), or we can revisit in the future.

[chandrarn]

COCOS conversion looks good, branch confirmed to run with TARS. Branch does not yet integrate into my one-equilibrium-reconstruction ChiSq fit code due to a difference in how the sensor names are interpreted (my branch still uses the C-Mod sensor naming convention), but this issue can be resolved on my end, or in a later commit.
Code is confirmed to work with gEQDSK files on my end, and produces a reasonable synthetic Mirnov signal output for a synthetic 1/1 mode.