"inline" system test fails intermittently

• https://gitlab.isc.org/isc-private/bind9/-/jobs/571680
• https://gitlab.isc.org/isc-projects/bind9/-/jobs/574812

I believe this is caused by a missing call to nextpart() before a call to wait_for_log(), which causes the latter to match the all zones loaded line for the previous named instance, effectively becoming a no-op (the wait_for_log() call in question is supposed to prevent invoking rndc zonestatus before all zones are loaded):

$ grep -E "(halt|zonestatus|all zones loaded)" bin/tests/system/inline/ns3/named.run
...
16-Jan-2020 15:45:52.765 received control channel command 'halt'
16-Jan-2020 15:45:54.007 all zones loaded
16-Jan-2020 15:45:54.045 received control channel command 'halt'
16-Jan-2020 15:45:55.295 received control channel command 'zonestatus delayedkeys'
16-Jan-2020 15:45:55.307 all zones loaded