Fix cache metrics to collect from all servers

[MadLittleMods]

Fix cache metrics to collect from all servers

Follow-up to #18604

Previously, our cache metrics did include the server_name label as expected but we were only seeing the last server being reported. This was caused because we would CACHE_METRIC_REGISTRY.register_hook(metric_name, metric.collect) where the metric_name only took into account the cache name so it would be overwritten every time we spawn a new server.

This PR updates the register logic to include the server_name so we have a hook for every cache on every server as expected.

I noticed this problem thanks to some tests in the Synapse Pro for Small Hosts repo that sanity check all metrics to ensure that we can see each metric includes data from multiple servers.

Testing strategy

1. This is only noticeable when you run multiple Synapse instances in the same process.
2. TODO

(see test_cache_metric_multiple_servers that was added in this PR)

Pull Request Checklist

• Pull request is based on the develop branch
• Pull request includes a changelog file. The entry should:
  • Be a short description of your change which makes sense to users. "Fixed a bug that prevented receiving messages from other servers." instead of "Moved X method from EventStore to EventWorkerStore.".
  • Use markdown where necessary, mostly for code blocks.
  • End with either a period (.) or an exclamation mark (!).
  • Start with a capital letter.
  • Feel free to credit yourself, by adding a sentence "Contributed by @github_username." or "Contributed by [Your Name]." to the end of the entry.
• Code style is correct (run the linters)

[MadLittleMods]

Removed because it was unused

[MadLittleMods]

Same changelog as #18604 so they merge

[MadLittleMods]

Thanks for the review @devonh 🐠