Toggled default for relay_log_recovery

In a replication failover scenario, we need to wait for the relay logs to
be applied on the follower, BEFORE promoting it to read-writable. This
involves a restart of the node, and relay_log_recovery was set to 1,
which meant we were blowing away relay_logs, which could be potential
loss of data.

So toggling the value, so that we have the relay logs, to try to
minimize data loss when failing over.

[#164675958]

Signed-off-by: Jamie Monserrate <jmonserrate@pivotal.io>

jobs/pxc-mysql/templates/my.cnf.erb

@@ -165,7 +165,7 @@ event_scheduler                 = <%= bool_to_on_off(p('engine_config.event_sche
 master_info_repository          = TABLE
 relay_log_info_repository       = TABLE
 relay_log                       = mysql-relay
-relay_log_recovery              = 1
+relay_log_recovery              = OFF
 master_verify_checksum          = 1
 
 <% if p('engine_config.audit_logs.enabled') %>