- even/odd is not consistently implemented in invert
  if there is one operator without we need to get the memory 
  for even_odd_flag == 0!!

- SourceSink_Pairs propagator format possibly broken for 2 fl. tmwilson
- residuum check for 2 fl nd twisted fails
- do we need basenames per operator
- introduce source types
- volume sources
- spin shifts in sources
- adaptive precision for overlap
- adapt online measuremnts for new operator structure
- incorporate stouting into new operator structure
- Remove the DUM_* variables and g_spinor_fields
- incorporate overlap operator
- append/not append feature for spinor fields
- source generation inside the code
- deflation
- mixed precision solver, single precision Dirac operator
- GPU implementation...
- test RHMC?
- dynamical overlap code
- create simpler test suite
