To calculate the performance impact of the ingress request rate limiter over a large sample, we should introduce a new buildvariant that has the rate limiter enabled with an effectively infinite rate limit. Initial sys-perf patches indicate there's some level of regression, and having the buildvariant run periodically over time will allow us to better quantify this regression (if any).

It should be sufficient to only duplicate a single topology (e.g. replset). This variant shouldn't be included in server engineer patch builds, but it should be run periodically to ensure the samples are being collected.