Issue 8

Created by: martukas

A modification of Doro's clustering algorithm is now used in the pipeline and works with both APV-originated data and VMM acquisition replays. It is not perfect, but definitely better tested and more robust. I have incorporated some of the comments from the code review. The ones I have not taken action on are left as TODOs for some later point in time when we have a better understanding on how we need to optimize it. We should merge the current state because: A) I am moving on to work more on multigrid, so there will likely not be much work on this branch in the near future B) it brings in proper use of the most recent h5cpp version, we can start adapting the raw data writing mechanism in other pipelines by example of gdgem C) The stuff you are doing in #46 (closed) will also be diverging from this, so let's touch base in master D) We need to get on the same page with the NBI guys, so let's make these changes official

Caveat: I did not build performance tests. They very likely no longer work. Until we do have a conan package for them and running on jenkins, I do not feel obliged to have them always working :)