Description of problem:
Currently in puppet-odl we are using an exec for a bash script to configure clustering which generates the the clustering config files for ODL (akka.conf, modules.conf, module-shards.conf).  We need to migrate away from a bash script and in doing so also provide a proper default configuration.  The bash script creates a default configuration with several modules/shards, including a dummy example module which should not be in production.

The new method will use erb templates to create the files, and by default place the entire datastore under just the default module.  Meaning the entire datastore will be led by a single ODL instance.  Optionally a user can use a new parameter to puppet-odl 'ha_db_modules' in order to break up the datastore into multiple modules/shards.