[Bug] Probably memory leak while publishing ROS topic continuously

[evshary]

Describe the bug

I ran the Autoware with AWSIM simulator, which they are in different domains and connected to each other with zenoh-bridge-dds.

 (domain 1)       (domain 1)              (domain 2)        (domain 2)
    AWSIM ---- (zenoh-bridge-dds) --- (zenoh-bridge-dds) --- Autoware

I found that memory usage was continuously increasing, and the zenoh-bridge-dds would crash.
Here is the video while memory usage increased

2023-02-15.zenoh-bridge-dds.memory.leak.Autoware.-.simplified.mp4

I guess the issue would happen while publishing data continuously.
To simplify the scenario, I ran two ROS nodes with zenoh-bridge-dds connected.
One of the nodes received image data from the camera and published it to another node.
I found that the same issue still happened.

To reproduce

1. Open 1st terminal and publish image from camera: ROS_DOMAIN_ID=1 ros2 run image_tools cam2image
2. Open 2nd terminal and subscribe image: ROS_DOMAIN_ID=2 ros2 run image_tools showimage
3. Open 3rd terminal and run the bridge: ./target/release/zenoh-bridge-dds -d 1
4. Open 4th terminal and run the bridge: ./target/release/zenoh-bridge-dds -d 2

Then using htop to observe the zenoh-bridge-dds memory usage.

System info

• Platform: Ubuntu 20.04
• ROS2 version: galactic
• zenoh-bridge-dds version: f94ae61

[JEnoch]

After analysis, the leak occurs when a data is routed from Zenoh to DDS, calling dds_writecdr().
The bug is in cyclocut. Fix to come soon.