refactor(server): replace stream() with subscribe() + startTurn() across all providers

Replace three competing event paths (foreground stream, live event pump,
JSONL history poller) with a single push-based subscribe() + startTurn()
contract. This fixes duplicate user messages and stuck running state caused
by timing-based routing between concurrent event sources.

Key changes:
- AgentSession interface: remove stream(), add subscribe() and startTurn()
- All providers (Claude, Codex, OpenCode): single subscribers Set with
  notifySubscribers() for push-based event delivery and turnId stamping
- Agent manager: identity-based turn ownership via activeForegroundTurnId
  replacing pendingRun async generator
- Delete: dual queues, routeSdkMessageFromPump, startLiveHistoryPolling,
  snapHistoryOffsetToEnd, liveEventBacklog, Pushable
- Fix Codex provider not clearing activeForegroundTurnId on turn completion
- Add real-provider integration tests for event stream invariants

packages/server/src/server/agent/mcp-server.ts

@@ -199,10 +199,7 @@ function startAgentRun(
   logger: Logger,
   options?: { replaceRunning?: boolean },
 ): void {
-  const snapshot = agentManager.getAgent(agentId);
-  const shouldReplace =
-    options?.replaceRunning &&
-    Boolean(snapshot && (snapshot.lifecycle === "running" || snapshot.pendingRun));
+  const shouldReplace = Boolean(options?.replaceRunning && agentManager.hasInFlightRun(agentId));
   const iterator = shouldReplace
     ? agentManager.replaceAgentRun(agentId, prompt)
     : agentManager.streamAgent(agentId, prompt);
@@ -696,7 +693,7 @@ export async function createAgentMcpServer(options: AgentMcpServerOptions): Prom
         throw new Error(`Agent ${agentId} not found`);
       }
 
-      if (snapshot.lifecycle === "running" || snapshot.pendingRun) {
+      if (agentManager.hasInFlightRun(agentId)) {
         waitTracker.cancel(agentId, "Agent run interrupted by new prompt");
       }