Not sure how useful P3 is. Out of curiosity I’ve attempted to run it a few times awhile back on my dev site … it was chock full of PHP notices and would never finish a scan because of of running out of memory, not matter how much I assigned to php. Just tried it a gain, same results, unfortunately. If I understand what the pie chart is really to represent, those results can be skewed one way or another with a bias towards any one plugin because it is created by sampling a set of pages. For example is a scan has more pages that had Connections on it than lets say Nextgen , Connections would take more average load. The same would be the opposite.
I have 81 plugins activated on this site, which includes Connections and every extension and template for it, and a typical admin page load uses between 50-60MB. I haven’t measured page load as I use W3TC to really optimize the page load times.
As for how to do memory leak checks, honestly don’t know.