Add alertmanager cluster flags to documentation

[bboreham]

If you run more than one alertmanager you need to have them talk to each other, governed by these flags from upstream: https://github.com/prometheus/alertmanager#high-availability

However I can't see that we explain this anywhere in the Cortex docs.