Fix race condition in get/get-config netconf rpcs for config subsystem

Fix race condition in get/get-config netconf rpcs for config subsystem

The read and transformation of modules from config subsystem is performed in a couple of steps. Each step queried the config subsystem on its own so it was possible to cause inconsistencies among these steps if get/get-config was executed while another transaction was being committed. This race condition, if hit, caused the rpcs to fail with an exception of a missing MBean.

After recent introduction of "optional reconnect after capability changed" feature into sal-netconf-connector, this race condition caused get rpc invoked by the connector to fail.
Scenario:
odl-netconf-connector-all is installed,
after the loopback is connected any other feature with configuration is installed e.g. odl-restconf-all,
loopback connector reconnects and as it comes back executes get rpc (to query ietf-netconf-monitoring)
get rpc in config-netconf-connector fails to transform the modules into xml due to mentioned race condition as the initial config file for restconf has been pushed

Now the get/get-config read the data from a dedicated transaction started just for the read.
Note: get rpc reads the runtime beans that are part of no transaction, so there is still possibility for this race condition regarding runtime beans.

Change-Id: I0822bc48745f9f680b116095693052dff752dee3
Signed-off-by: Maros Marsalek <mmarsale@cisco.com>