should transfer leader concurrently when update configuration

[HarrisChu]

when update storage configuration, it would be rolling update the storage pod.
but transfer the leader part by part, and it would be a long time for some queries, e.g.

match (v) return v limit 10

expected:

1. transfer leader concurrently.

[kqzh]

I try to transfer leader concurrently, but get many errors cc @HarrisChu @MegaByte875

[HarrisChu]

please check if multiple go routines use the same tcp connection.

[MegaByte875]

#255

[jinyingsunny]

development students selftest，sc not read from leader.