New Comparison View: Make sure we do not use timing values from profiling runs, which may be influenced by profilers.

[smarr]

With the combining of data from different experiments and trials, there is a risk that we use the data from profiling runs to compute statistics.
But, these might be influence by the profiler, so, it would be better not to use them.

This is possibly an issue since or exacerbated by #164.
It might not be one in practice, depending on how the profiling is set up.
But still worthwhile to verify.