Consider sending more traffic to registry cny stage

(I tried looking for an issue about this, but couldn't find one, please feel free to close as a duplicate though)

Currently, we send ~0.2rps through registry/cny. This is very low and results in dirty, low-volume metrics on our Registry SLIs:

https://dashboards.gitlab.net/d/registry-main/registry-overview?orgId=1&var-PROMETHEUS_DS=Global&var-environment=gprd&var-stage=cny&var-sigma=2&from=now-7d&to=now&fullscreen&panelId=3

Since we use the same thresholds for cny and main stages in our service monitoring, this noise is preventing us from tightening up the SLAs for the registry service (which, on the main stage, has excellent metrics these days).

Sending more traffic this way would certainly help.

cc @jarv @skarbek