It's back: Timeout (XXs) waiting for privilege escalation prompt

This problem is back, and it is worse than ever in 2.6.1, from epel, on el7.5.

If I run a playbook (just one or two simple tasks) vs 50 hosts, I’ll have a dozen that fail due to timeout…

When I run the same playbook vs same hosts from server with ansible-2.4.2.0-2.el7 … it works perfectly.