Comment (by Zach Etienne):

Hi @Ian Hinder . My patch for corrected midpoint is just above your message. Probably needs a bit more work ito documentation, but it does the job. @Roland Haas decided to change the default to trapezoidal for the release, since (1) this is not a midpoint method at all (as you know), (2) it produced very wrong results (~few% error on BBH waveform extraction, unnecessarily), and (3) the bug was uncovered too close to the release.

--
Ticket URL: https://bitbucket.org/einsteintoolkit/tickets/issues/2398/multipole-default-midpoint-integration