We (the authors of the plugin, i.e. Diabol), believe that manual triggers DO make sense in a pipeline. A high degree of automation is great and necessary in a pipeline, but some decisions will be taken by humans in many pipelines. What should always be automated is the repetitive tasks. This is one of the reasons that we believe that the total CYCLE time is an important measure.
Also, I think your presentation is focusing too much on 'anecdotal' measurements. It's the statistical values that should matter. I attach a suggestion for displaying total cycle time and total build time. For the aggregate row, it's the same but with statistical values: average, minimum, median, 90-percentile and maximum. How about that?
Thank you for the screenshots associated with the pull request.
I would prefer to display the actual time for the pipeline to finish, rather than just adding up the build task duration for each task. In a pipeline with parallel stages the figure would otherwise be deceptive.