[FIELDLINES] allow start from plasma boundary given in VMEC INDATA, not only wout

[jonathanschilling]

These changes allow to provide the command-line argument -edge to FIELDLINES not only in combination with tracing over a VMEC equilibrium, but also in vacuum (-vac).
The plasma boundary geometry is then read from the VMEC INDATA namelist using RBC and ZBS (and in case of lasym, also RBS and ZBC). Furthermore, nfp, mpol and ntor are needed.

Additional input parameters (in the FIELDLINES_INPUT namelist) named nstart_pol and nstart_tor are provided to specify how many grid points over the plasma boundary are to be used in each of the tangential directions. They default to 0, in which case (as if they are not specified in the FIELDLINES_INPUT namelist) operation defaults to int(floor(sqrt(maxlines))) per direction, as previously implemented.

[jonathanschilling]

Note: Assuming one has an existing wout file, but only wants to use the plasma boundary geometry from there, one can copy that plasma boundary geometry into a new VMEC indata namelist using the wout2indata tool from #245.

[jons-pf]

@lazersos FYI, I think this should now be compatible again with the recently merged py_stel update.

[lazersos]

These should only be written out if they're >0.

[lazersos]

The modifications in FIELDLINES_INIT_MGRID and COIL (All the BCASTS) shouldn't be there. When you call init_edgestart, you should re-read the INDATA namelist. This better compartmentalizes the code. Also mylocalmaster doesn't need to be broadcast.