Documentation request: details on sidecar instance scaling and traffic routing

@damemi and @ymotongpoo, thank you **so much** for this example repo and it's a thrill to see the sidecar feature launched and available! 🙌 👏

I’m interested in using cloud run for precisely this use case: adding an OpenTelemetry collector to a cloud run service with several dozen container instances but after an early experiment based on this repo and reading the docs I’m unclear on two points:

1️⃣ How do sidecar container instances scale relative to the main app container?
- In deploying an example project, logs showed multiple sidecar instances launched as traffic increased to the main service. Is this scaling based on resource use (cpu/memory) for the sidecar instance, or on something else?
- Are there any configuration options available to control this? For example if we wanted to run just a single OpenTelemetry collector alongside a large number of app instances, could we provision a large amount of resources for the collector instance to prevent additional instances from launching? Would a different configuration or architectural pattern be better suited?

2️⃣ When multiple sidecar instances are running how is traffic routed to them? Is it load balanced across instances?
- For example, if multiple OpenTelemetry sidecar instances are launched and all listening on the same port, will traffic make it to each instance?

Thanks again for the clear and practical examples and if these questions are already answered in documentation I’ve missed or would be better directed at a Cloud Run product team, please advise!