Planned improvements to current diffusion

[bclyons12]

An evolving list of things that should be done for the current diffusion solver. Please check off as completed and post a comment with what was done and appropriate commits/PRs.

• [x] Add non-inductive current source
• [x] Steady-state diffusion
• [x] 
Add other boundary conditions
  • [x] Constant Ip

• [x] Complete thorough benchmarks
  • [ ] TRANSP
  • [ ] ONETWO
• [x] Optimize code, especially:
  • [ ] matrix-computation routines (define_T and define_Y)
  • [ ] inner_product integrals
• [ ] Allow for arbitrary grid in diffuse, including upsampling and non-uniform packing
• [ ] Create more initialization routines
• [x] Make accessible through OMFIT

I could use @lstagner and @orso82 keen eyes on the optimization. I'm sure there's a lot of low-hanging fruit.

[orso82]

@bclyons12 could we have a way to estimate a good time-step based on the values of the resistivity profile, and use that to set the number of time-steps given a tmax?