Rebalance failure due to replicator_died with "memcached_error,auth_error,<<"Auth failure">>" because node was marked as ready before isasl.pw was read

Decided to file a new bug, but the old one is here:-
http://www.couchbase.com/issues/browse/MB-5259
http://www.couchbase.com/issues/browse/MB-5291

<0.4181.3> exited with {exited,
{'EXIT',<0.4182.3>,
{replicator_died,
{'EXIT',<15697.23073.1>,
{badmatch,
{memcached_error,auth_error,
<<"Auth failure">>}}}}}}

Failing testcase:-
swaprebalance.SwapRebalanceFailedTests.test_failover_swap_rebalance,replica=1,num-buckets=4,num-swap=1,swap-orchestrator=True
1) Failover a node orchestrator
2) Swap rebalance it.