Title: #956: tests: fix race confition in enumeration tests
In other words, you think it is safer to have INTERACTIVE_TIMEOUT*2
where the previous /2 was?
Actually I liked initial Pavel's approach (INTERACTIVE_TIMEOUT=2*ENUMERATION_TIMEOUT)
more. Because we need to have all sleep_timeouts > enumeration_timeout. Not just in the
beginnings of a test.
What was missing: replacement INTERACTIVE_TIMEOUT/2 with INTERACTIVE_TIMEOUT in the
beginnings (this is what Fabiano's patch does).
Those patches combined plus your explanation of reason behind first sleep() seems
reasonable to me.
See the full comment at https://github.com/SSSD/sssd/pull/956#issuecomment-562256253