Move plotdata call earlier in eikcoefs to avoid issue with theta being redefined.

For certain equilibrium types (at least Efit and GS2D) the Rpos(r, theta) and Zpos(r, theta) functions require the theta passed to be consistently defined with that used to calculate the values of r passed in (from rgrid). Here we move the call to plotdata, which calls Rpos and Zpos to before the handling of equal_arc as this could lead to a redefinition of theta.

This fixes bugs in the netcdf output of Rplot, Zplot etc. but I don't think it should impact on the outputs of geometry used in the rest of GS2.