output time of vertex duration in run statistics

[ManuelHu]

• Outstanding question: Should we include the time of the first event geometry initialization in the statistics? Currently it is included in the time of the first vertex generation, which significantly skews the statistics for the first run...
• This also reset the trial counter (and now the generation time) before each new run, so that it does not monotonically increase which each run (as it did until now)

Other changes:

• Fixes allowed states for all confinement commands
• Validate parameters for geometrical solids (solid types, all parameters). Now one cannot modify a box as a cylinder :-)
• Move geometrical solid type from string to an enum (the strings did not match G4 names anyway)
• Add a command to reset the vertex confinement parameters. This should allow to reset+reconfigure confinement between runs

fixes #48