Feature: crmd: update node state for started pacemaker_remote nodes

When a pacemaker_remote node first starts, the connection host initializes its
node attributes and CIB state. Previously, these were done at two different
points in the code. Now, they are both done when the remote connection result
is reported, and additionally the host notifies the DC via a cluster message.

This helps keep remote-specific code in remote_lrmd_ra.c,
eliminates the DC needing to check operations for remote node effects,
and reduces code duplication by using do_update_node_cib() instead
of needing a separate simple_remote_node_status() function.

This fixes one bug: Previously, if a remote connection did not have
a reconnect-interval configured, and the node was stopped or lost and
later rejoined, the PE would not be invoked immediately, so resource placement
would not occur on the rejoined node. Now, the DC state message will
cause the peer update callback to be called, which will abort the transition,
triggering resource placement.