[Backport 10.2] RestProxyService no longer works when running a multi-node cluster on the same server

The CMS RestProxyService is used for backchannel REST communication with the ChannelManager.
The rest.uri which is used for this backchannel communication is configured as property in the repository on node:

/hippo:configuration/hippo:frontend/cms/cms-services/hstRestProxyService

and by default bootstrapped as:

rest.uri = http://127.0.0.1:8080/site/_cmsrest

This also means that for a clustered setup each instance will have to use the same value.

This works fine, as long as cluster instances are hosted by different servers but are running on the same port.
But this breaks when multiple instances are hosted by a single server on different ports, as that will direct all the RestProxyService communication to only one of these.
That is no longer technically possible, nor desirable, since the improvements introduced with CMS-10424.

Being able to run a multiple Hippo instance cluster on a single server, while maybe not a desired nor likely production setup, can be convenient for test and development purpose.

To allow such a setup, we will allow the rest.uri to be specified without port, e.g. like:

rest.uri = http://127.0.0.1/site/_cmsrest

If no port is specified, the RestProxyService will use the current CMS request port also for the REST backend communication with the ChannelManager.