Title: #947: tests: fix race conditions in integration tests
@pbrezina, Hi Pavel :)
I don't really remember, sorry, but looking at the code now it might be to avoid
racing with SSSD initialization, while making the change before the timeout expires. I.e.
to have the database change between the SSSD initialization and the first timeout,
If value of this timeout has nothing to do with enumeration period I would prefer to
replace/rename `INTERACTIVE_TIMEOUT/2` as this is a little bit misleading.
See the full comment at https://github.com/SSSD/sssd/pull/947#issuecomment-559488780