docs: handling of raft_multiplier in a production cluster is not documented

There are two issues on which documentation is clearly missing:

1. What will happen if consul servers with different raft_multiplier settings are joined into a cluster? Which setting will win, the quickest, the slowest, the oldest, the newest... ?
2. How to change the value of raft_multiplier if the cluster is already in production? Is restarting each member with a new value required or can this be changed on live nodes?

No file or test is named in the issue. Start by locating the existing Consul documentation for raft_multiplier and cluster membership, then establish the behavior for mixed settings and the supported production change procedure. Done means documenting clear answers to both questions, including any restart requirements.

Written by the indexing model from the issue text.