Webhook driven scaling questions

[wr37ch]

We're currently exploring the transition from PercentageRunnersBusy to webhook-based scaling, but I have a couple of questions. I'd appreciate some clarification on the following:

1. Rate Limiting and Webhook Driven Scaling

On the documentation page, it mentions,

If you are using controller version < v0.22.0 and you are not using GHES, and so you can't set your rate limit budget, it is recommended that you use 100 replicas or fewer to prevent being rate limited.

Does this advice also apply to webhook-driven scaling, or is it specific to pull-driven scaling, is opting for webhook-based scaling a good idea when you're running low on rate limits? Also, is there a way to estimate the number of requests someone would need to make to the GitHub API?

2. Scaling Behavior with Webhook-Driven Scaling

During the testing, I observed a scaling behavior that caught me off guard. I initially assumed that with webhook-driven scaling, and no connectivity issues, runners would be spun up by a webhook and terminated as soon as the job finishes. However, it seems that the HRA sets a desired number of runners, likely based on the load. Could you provide insights into how it determines the desired number of runners at any given moment?

Thanks.