Add latency metrics to eviction pools routines

[yohkaz]

(potentially) Closes #6664

[mattsse]

these calls are very time sensitive

to me the additional clock calls seem alright, wdyt @hai-rise

[hai-rise]

@mattsse I think they are alright, too.

This won't likely affect us as we can buff the mempool size of the centralized sequencer and buffer new transactions instead of truncating when full 🙏.

[emhane]

would be nice if pr included update grafana dashboard code to use these metrics + a screenshot of how it looks

[yohkaz]

@emhane I followed this guide to add the new metrics to grafana dashboard: https://github.com/paradigmxyz/reth/blob/main/etc/README.md
for some reason grafana decided to move around existing metrics in the changeset: 56358d4
I think it fixes some position issues of existing panels on the way.
If you know how to avoid this, lmk.

This is how the new metrics look:

[emhane]

lgtm. grafana dashboards are tricky to update, they always leave a lot of diff, that's why we separated the dashboard into several different dashboards e.g. one for mempool, one for discovery, etc.

[yohkaz]

@mattsse @hai-rise is there any other concern/request?