Currently, the connection sys-perf workloads are only tracked by the default performance contexts, namely "priority1" and "majority lag" (e.g. this). The "priority1" context only tracks generic metrics such as operation latency or throughput per user/phase. We should define a performance context for connection metrics, such as "connections.created", "connections.current" and "connections.active", so perf BFs are created automatically when there is a regression/change point in the connection code paths.