Investigate difference in results when using newer OpenCV version

When using a newer OpenCV version than 4.2. leads to failing reference tests, meaning that a significant difference in the results occurs. This prevents us from moving to a newer version of OpenCV which makes it difficult on some systems to compile PeTrack without compiling OpenCV. In particular some package managers to have this particular OpenCV version , e.g, pacman, homebrew.

Following should be done:

• run PeTrack for multicolor and aruco TdN test data sets
• if possible find the origin of the differences
• find a measure to judge results: Are the results getting better or worse with newer OpenCV version?
• decide if a update to a newer OpenCV version is possible

This issue can be used to store the information from the different steps.