Are you sure that the current test only runs for a few seconds? It is hard for me to believe with 1000 timesteps and 1100 particles.
Did you test the quintet-stuff with more than one species WITHIN the quintets?
Can you put such a test into the testsuite and your test with a single species in the membrane into the examples? Sorry to ask for that, but multi-species-tests are critical. Many bugs only show up in such a case. Take care how particles are indexed in the connector file in such a case. It goes like (0..9) for the 10 particles belonging to the first species and then (10..14) for the 5 particles belonging to the second species. Ideally, you extend your current test by "simply" changing the species of some particles of your choice in the quintets. The interactions in the quintet and with the fluid may remain identical, except that you now have to define them for additional species-interactions. In this way you should get exactly the same output (if kinetic energy of all species in the membrane is properly measured and summed) and can easily compare for correctness. I know, this is a bit of work, but necessary for a reliable code.