You are able to deploy 100 times every 86400 seconds (1 day). Should you hit the rate limit, you will need to wait another day before you can deploy again.
It is every 86400 seconds window, so ideally should reset 1 day after the first request in the previous window. This means, say you deployed 100 times in last one hour, then you need to wait 23 more hours irrespective of the time of the day.
You should get rate limit headers API responses, which indicate the usage, remaining, and reset time.
Yes. All deployments no matter what caused them (API or Git event) will count in the same quota.
Can you share the use case and workflow you’ve? This will help us understand why you might be hitting the limits and what can we do to avoid.