In the RestRequestCache, we are trying to limit non GET calls for a specific host and user to one call per second (and all calls, including GET, for a specific host and user, need to be serialized). This throttling is based on recommendations from GitHub (https://docs.github.com/en/rest/guides/best-practices-for-integrators?apiVersion=2022-11-28#dealing-with-abuse-rate-limits). The current implementation isn't really doing that however. Calls are serialized, but depending on timing, we can definitely send more than one non GET call per second, at least in bursts.

I think that if we really want to enforce such a limit, we need to rework this and use two different locks, one for all remote calls and an additional one for non-GET calls. The GET calls would simply look like this (as they do today):

1. Take lock for general remote call
2. Perform GET remote call
3. Release lock for general remote call

For the non-GET remote calls, we would have this:

1. Take lock for non-GET call
2. Wait if we need to (compare timestamp with timestamp from last call)
3. Take lock for general remote call
4. Update timestamp for last non-GET call
5. Perform non-GET call
6. Release lock for general remote call
7. Release lock for non-GET call