Parse command-line parameters with boost program options

[slizzered]

Boost program options ftw!

This will also define a syntax for a possible config file, see #10

Multiple steps are necessary:

• [x] use CMake (see #27)
• [x] fix the CPLUS_INCLUDE_PATH (see #26)
• [x] fix compiling issues related to the Boost 1.55-fix (see #19)
• [x] think deeply about the structure of the parameters:
  • should there be switches? (like --use-reflections to enable reflections)
  • should there be only options with defaults? (like --use-reflections=1 to enable reflections)
  • what is a good way to specify the runmodes
  • what parameters must be given every time?
  • can we use sane defaults for everything else?
• [x] with this new knowledge, implement the behaviour

[slizzered]

development of this feature can be tracked here for now: https://github.com/slizzered/haseongpu/commits/issue15-boost_program_options

[slizzered]

current proposal: (short forms are only for the more common options) good short forms for --min-rays and --max-rays would be needed.....

Long	Short	Use	Meaning
--help	-h		print the help
--verbosity	-v	--verbosity=	define the verbosity (has a default)
--device-mode		--device-mode=(cpu|gpu)	defines on which hardware the calculation is performed
--parallel-mode		--parallel-mode=(mpi|threaded)	defines if threaded mode or MPI mode us used *
--reflection		--reflection=(1|0)	use reflections or not *
--min-rays		--min-rays=N	the minimum number of rays for the adaptive steps
--max-rays		--max-rays=N	the maximum number of rays for the adaptive steps *
--input-path	-i	--input-path=	the location of the input files (defaults to ./input)
--output-path	-o	--output-path=	the location of the output files (defaults to ./output)
--ngpus	-g	--ngpus=N	the number of GPUs to use (can be less, if less are available. This will print a warning. Defaults to 1 *
--min-sample-i		--min-sample-i=N	the index of the first sample point to process (in a range to --max-sample-i). Defaults to 0 *
--max-sample-i		--max-sample-i=N	the index of the last sample point to process (in a range from --min-sample-i). Defaults to the maximal index of the mesh *
--mse-threshold	-m	--mse-threshold=N	The MSE threshold that is used to stop adaptive sampling (defaults to 0.1, which is pretty high) *
--spectral-resolution		--spectral-resolution=N	The number of points used to interpolate the intensities for the laser spectrum. N must be at least as high as the number of data points for the spectrum. Defaults to the number of data points for the spectrum *
--repetitions	-r	--repetitions=N	The number of retries for repetitive sampling (defaults to 4, which is a good default) *
--adaptive-steps	-a	--adaptive-steps=N	THIS IS NOT TO BE IMPLEMENTED NOW *

*\ This option is not** valid in combination with --device-mode=cpu

--write-vtk and --compare are cancelled for now.

Any constructive criticism is welcome

[erikzenker]

Very good, would do it the same :ghost: