The former pertains to discovery: it allows you to specify the addresses of the master nodes. This is needed whenever a node starts up, or a new master is elected. The latter pertains to cluster bootstrapping and requires the identities (i.e. the names) of the master nodes. It only happens once in the lifetime of a cluster. They're very different things. The older minimum_master_nodes mechanism was removed a very long time ago since it was fundamentally unsafe and led to data loss.
This is covered by the docs linked from the ones you linked:
If you leave
cluster.initial_master_nodesin place once the cluster has formed then there is a risk that a future misconfiguration may result in bootstrapping a new cluster alongside your existing cluster. It may not be possible to recover from this situation without losing data.