Documentation: alert for grpc related methods is outdated

[dqminh]

With commit 5e810e3, etcd defers to go-gprc-prometheus to set up prometheus metrics.

The new prometheus metrics are like:

grpc_server_handled_total{grpc_code="OK", grpc_type="unary"}

But alerts refenrence guide is still using old metrics:

IF sum by(grpc_method) (rate(etcd_grpc_requests_failed_total{job="etcd"}[5m]))
  / sum by(grpc_method) (rate(etcd_grpc_total{job="etcd"}[5m])) > 0.01

IF histogram_quantile(0.99, rate(etcd_grpc_unary_requests_duration_seconds_bucket[5m])) > 0.15

We should update the reference guide to use new prometheus metrics instead.

cc @gyuho @brancz

[gyuho]

Closing via #8879.