feat(swarm): heartbeat-driven topology resync for agent-pinned deployments

Agent heartbeats now carry an applied-topology snapshot. The master
heartbeat handler compares the reported version_hash against what
canonical_hash yields for the hydrated topology pinned to that host
and flags Topology.needs_resync on divergence (or when the agent
reports no topology at all while master expects one).

The mutator watch loop gains reconcile_agent_resyncs, which re-pushes
the current hydrated blob via AgentClient.apply_topology without
touching status, then clears the flag on success. Push failures leave
the flag set so the next tick retries.

decnet/web/db/models.py

@@ -225,6 +225,11 @@ class Topology(SQLModel, table=True):
     # the topology or any child row when an expected_version is supplied.
     # Callers pass their last-seen version; mismatch raises VersionConflict.
     version: int = Field(default=1, nullable=False)
+    # Set by the heartbeat handler when an agent's reported
+    # ``applied_version_hash`` diverges from what we expect it to be
+    # running.  Drained by the mutator watch loop, which re-pushes via
+    # AgentClient and clears the flag.  NULL for unihost topologies.
+    needs_resync: bool = Field(default=False, nullable=False)
 
 
 class LAN(SQLModel, table=True):