Explain that cockroach init must target the node(s) mentioned in --join lists

[knz]

See this issue: cockroachdb/cockroach#61621

If there are multiple nodes connecting together,

then an init command pointed to a node that's not listed in the --join list of other nodes will fail to initialize the cluster correctly.

The doc should have a call out that init should target a node that's listed in the --join list of all the other nodes.

(Or alternatively, a cluster can be initialized with just 1 node, and the other nodes added after initialization. In that case, the join flags are not restricted.)

cc @taroface