Fix routing loop when 2 Subscribers for a same topic exist across 2 bridges

[JEnoch]

In a setup with 2 interconnected bridges, with for instance one on domain 0 and the other one on domain 1:
if in domain 0 there are a Publisher and a Subscriber on a same topic T and in domain 1 a Subscriber on topic T, then the Subscriber in domain 1 receives twice each publication.

E.g. to reproduce on a single host:

• ROS_DOMAIN_ID=0 ros2 run demo_nodes_cpp talker
• ROS_DOMAIN_ID=0 ros2 run demo_nodes_cpp listener
• ROS_DOMAIN_ID=0 zenoh-bridge-ros2dds
• ROS_DOMAIN_ID=1 zenoh-bridge-ros2dds
• ROS_DOMAIN_ID=1 ros2 run demo_nodes_cpp listener

The problem comes from bridge on domain 1 that re-publishes to DDS the data coming from bridge on domain 0, but also receives back this publication and route it back to the bridge on domain 0.
The cause is the DDS Writers (and DDS Readers) created by the bridge on a remote announcement of route copy the same declared QoS without adding the ignore_local(Participant) QoS. Adding this QoS will make the bridge to ignore its own publications over DDS, avoiding to route them back.

This PR fixes that.

[JEnoch]

Note: the 2nd commit adds some cosmetic improvements in logs and comments

[Charles-Schleich]

Small nit:
do_route_message is an improvement over do_route_data
Not urgent or majorly important but clarity of direction of do_route_message functions could be improved.

Suggestion:
For route_subscriber::do_route_message and
route_publisher::do_route_message

It could improve clarity where as to direction to name the two functions differently.
But i suppose because they are in differently named source files, it should be fine.

Otherwise LGTM

[JEnoch]

Agreed. I'm renamed as such for clarity:

• route_subscriber::do_route_message => route_subscriber::route_zenoh_message_to_dds
• route_publisher::do_route_message => route_publisher::route_dds_message_to_zenoh