Revise automatic complexity measurements tests and reporting

There are a number of things to be checked in the automatic complexity measurements:

• Since the one-week outage of scheduled test runs, the ROM number graph is blank
• There can be failed modes, but no indication for that is given. This is somewhat by design to not have crashes in some modes block the reporting of all other modes, but there should be at least some notification
• There are some modes e.g. in stereo, where the bar plot for "Worst Case WMOPS Performance per Operating Point" does report zero, even though there are numbers in the corresponding log file
• Ideally measure for all selection test operating points to make sure we don't diverge from the release version too much