add display example image from zenoh

[jpizarrom]

This PR add display image example from zenoh for the uprotocol-sensors

[jpizarrom]

Hi @PLeVasseur

with this i show the images at 12/14 FPS on my laptop.
It looks like serde_json deserialize takes some time. if I don't do any operation/deserialize, only receiving the events, and just show a black image, got 35/40 FPS.

I did tried a bit to do the same on https://github.com/Eclipse-SDV-Hackathon-Chapter-Three/ego-vehicle-sensor-mock but just got 2/4 FPS

with python zenoh i was able to get 3/4 FPS, will try to do another PR with about this

[jpizarrom]

@PLeVasseur if i send pub array: Vec<u8>, // Flat RGB array: shape (height, width*4) from sensors, instead of pub array: ArrayView2<'a, Color>, i got 30/35 FPS display on Rust zenoh image display

i just started with Rust/Carla/SDV stack in the hackathon -> so not sure if i am optimizing in the right place

carla-zenoh-uprotocol-flat-vec.mp4

[jpizarrom]

@PLeVasseur i just updated this PR sample to serialize images as Vec<u8>, and this made Rust sample go from 10FPS to 35FPS, and also made Python client fluid. The json parsing was a bottleneck, for both Rust and Python, but affected much more my Python experiment.

the video shows CARLA Image over uProtocol over Zenoh for Rust and Python clients, both are fluid and probably with this changes, a ML model can use images over uProtocol over Zenoh to drive as POC. maybe Protobuf is better than just a vector, but nested dictionaries, can be a bottleneck.

the changes i did, are not zero-memory copy, i was more focused on the serialization.
I hope my conclusions are correct, as i just started with Rust, SDV stack, etc for the hackathon last week.

carla-zenoh-uprotocol-flat-vec-rust-and-python.mp4