Initial aether repository structure

Distributed actor system with event sourcing for Go:
- event.go - Event, ActorSnapshot, EventStore interface
- eventbus.go - EventBus, EventBroadcaster for pub/sub
- nats_eventbus.go - NATS-backed cross-node event broadcasting
- store/ - InMemoryEventStore (testing), JetStreamEventStore (production)
- cluster/ - Node discovery, leader election, shard distribution
- model/ - EventStorming model types

Extracted from arcadia as open-source infrastructure component.

Co-Authored-By: Claude <noreply@anthropic.com>

cluster/cluster.go

@@ -0,0 +1,48 @@
+// Package cluster provides distributed computing capabilities for the Aether VM runtime.
+//
+// This package implements a distributed actor system using NATS for coordination,
+// featuring consistent hashing for shard distribution, leader election for
+// coordination, and fault-tolerant actor migration between nodes.
+//
+// Key Components:
+//
+// - ConsistentHashRing: Distributes actors across cluster nodes using consistent hashing
+// - LeaderElection: NATS-based leader election with lease-based coordination
+// - ClusterManager: Coordinates distributed operations and shard rebalancing
+// - NodeDiscovery: Manages cluster membership and node health monitoring
+// - ShardManager: Handles actor placement and distribution across shards
+// - DistributedVM: Main entry point for distributed VM cluster operations
+//
+// Usage:
+//
+//	// Create a distributed VM node
+//	distributedVM, err := cluster.NewDistributedVM("node-1", []string{"nats://localhost:4222"}, localRuntime)
+//	if err != nil {
+//		log.Fatal(err)
+//	}
+//
+//	// Start the cluster node
+//	if err := distributedVM.Start(); err != nil {
+//		log.Fatal(err)
+//	}
+//
+//	// Load a model across the cluster
+//	if err := distributedVM.LoadModel(eventStormingModel); err != nil {
+//		log.Fatal(err)
+//	}
+//
+// Architecture:
+//
+// The cluster package implements a distributed actor system where each node
+// runs a local VM runtime and coordinates with other nodes through NATS.
+// Actors are sharded across nodes using consistent hashing, and the system
+// supports dynamic rebalancing when nodes join or leave the cluster.
+//
+// Fault Tolerance:
+//
+// - Automatic node failure detection through heartbeat monitoring
+// - Leader election ensures coordination continues despite node failures
+// - Actor migration allows rebalancing when cluster topology changes
+// - Graceful shutdown with proper resource cleanup
+//
+package cluster