v0.4

• added multi channel MC integrator with example
• added multi channel MC integrator with example
• added a new MC helper function hep::multi_channel_max_difference that
  calculates the maximum difference for adaptive optimization in multi channel
  MC
• added callbacks for multi channel MC analogous to the case for VEGAS
• added MPI parallelized routines for multi channel MC
• added new internal class hep::kahan_accumulator. This class takes care of
  the kahan summation
• changed interfaces mc_result and vegas_iteration_result: Their member
  variables are now properly encapsulated and accessible via public member
  functions. Furthermore, the mc_result now allows to obtain sum(),
  sum_of_squares(), and variance()