Provide a proper :recover management operation for transactions

Looking at the operator work for https://issues.jboss.org/browse/EAP7-1192[1], it seems that the operations required by the admin to perform recovery are quite complex.
It seems that the customer must to:

1. make sure that tx recovery listener is enabled in the tx subsystem
1.a read the correct port from the socket binding, don't forget to also read the portOffset and take it into account
2. connect to this tx-recovery-port opened by the server and send a SCAN command
3. read the Pod log with a regexp "ERROR.*Periodic Recovery"
3.a don't forget to timestamp the logs
4. read recursively all tx logs to check that there are no in-doubts tx

Oh and if you want to configure the tx recovery process, please add the correct system properties om.arjuna.ats.arjuna.common.Recovery*

This creates a lot of complexities for the customer (and the operator code) that should be encapsulated by the transactions subsystem.

Tthe subsystem should provide a :recover management operation
that deals with connecting with Narayana (no need to open a socket since Narayana is IN-VM) and perform the scan / logs checking / etc. and return the useful information to the user (the in-doubt TX)
Likewise, any recovery configuration should be handled by the transactions and not through System properties.

[1] https://github.com/wildfly/wildfly-operator/blob/master/pkg/controller/wildflyserver/wildflyserver_controller.go#L946