The problem went away when I trashed 3.6RC and dug 3.5 out of my TimeMachine backup.
at least to me, it doesn't make too much sense that the Qt5 launcher should have any effect on FG at run-time, due to the way it is integrated currently (or rather NOT integrated...)
Either way, I find it unfortunate to see that people have the tendency to revert back to older fgfs versions instead of following up with more useful information in order to troubleshoot/understand such issues properly.
No offense, but this is exactly how we arrived at having so many significant bugs with the only "solution" being to either delete $FG_HOME contents (e.g. the navcache) or even re-install FG from scratch (possibly even an outdated version).
Basically, that is the mentality of Windows users, and it is unfortunately not constructive to identify, understand and fix actual bugs - to see for yourself, just run a forum search for "sqlite" or "navcache/navdb", and you will literally find hundreds of end-users affected by features that didn't receive sufficient testing.
If we are now also going to see a similar degree of issues related to an OPTIONAL feature like the Qt5 dependency, we are going to cripple FG even more, which means that even fewer people will be able to use FG without resorting to weird workarounds in the form of having to delete/re-install stuff, just because there is so little testing/constructive feedback.