SingleFeatureTest fails to discover blueprint related problems for features which do not depend on the controller feature with the ODL specific blueprint extender

While fixing INFRAUTILS-10 I've realized that SingleFeatureTest does not actually discover blueprint related problems... Or at the very least it can clearly miss some of them.

For example, in that bug there was a BP XML with an invalid ref in a <service> and it missed that.. on looking more closely at the log, I believe what is happening is that TestBundleDiag currently only tests that bundles were installed and are "active", but may be missing to explicitly wait for BP validation?

What's curious is that SOMETIMES it hits them (which is was INFRAUTILS-10 was all about; in https://git.opendaylight.org/gerrit/#/c/56611/ the problem was found, but that was in SFT of integration/distribution ... which technically is the same code - but I suspect it may be slower there, and thus hit it? I think we have to await the same even in faster running individual feature tests, somehow.