+ slonczewski + oersted field now supported for all tp-schemes
+ oersted field is not recomputed if je = const(in time)

+ not finished
+ done only for TPS1
+ introduced ExplicitTerms = External + Oersted

+ magnetization was not updated correctly

+ only for tps1

+ not clear yet, what mathematical j_cf is exactly for Slonczewski / ZhangLi
+ wait for Michele's input
+ special care has to be taken for hc(v), hc(mOld) if current not time-dependent to be more efficient

Structure:
_magnetostaticField.py: Base class
  - {Stray,Oersted}Field derive from it
_detailsMagnetostaticField.py: static class variables for BEM stuff
  - K operator
  - BEM mesh
  - bemMesh <-> ngMesh mapping

{stray,oersted}Field.py:
  - Derive from _magnetostaticField.py
_details{Stray,Oersted}Field.py: static class variable for FEM stuff
  - forms, preconditioners, solver

- i.e. for strayField, oerstedField or magnetostaticField(=both)
- moved ComputeMaxwell to Magnetization class
- added numthreads to the test files
- added some missing import sys

+ moved all imports from top of files to functions/methods using them
+ using os.environ from within python to set environment variables for interpreter
+ use by: I = Integrator(..., numthreads=42)

+ bempp done
+ ngsolve done
+ omp not done

moved TOL, theta, preconditionerType, guessType to parameters
set up academic test example explaining everything

Reuse TestFunction(), TrialFunction(), Deriv() and use NgSolve's Vector / Matrix-Operations for VectorH1 functions and cfs