[FEATURE] Implement graceful shutdown for TrafficLayer

[yunlishao]

Is your feature request related to a problem?
Yes. Currently, shutting down TrafficLayer causes errors depending on the shutdown order:

• Close TrafficLayer.exe first → SUMO pops error (abrupt socket disconnection via libtraci)
• Close SUMO first → TrafficLayer pops error (socket error when trying to communicate)

Users need to know a specific shutdown order to avoid error popups, which is not user-friendly.

Describe the solution you'd like
Implement graceful shutdown that handles both shutdown scenarios without errors:

1. Dedicated cleanup function performCleanup(bool emergencyShutdown) with proper sequence:

   • Notify clients (send shutdown signal with state=0)
   • Call libtraci::Simulation::close() to close SUMO connection gracefully
   • Call socketShutdown() to close sockets

2. Enhanced Ctrl handler to catch all Windows console events:

   • CTRL_C_EVENT - Ctrl+C pressed
   • CTRL_BREAK_EVENT - Ctrl+Break pressed
   • CTRL_CLOSE_EVENT - Console window closed (X button) ← Currently missing!
   • CTRL_LOGOFF_EVENT - User logs off
   • CTRL_SHUTDOWN_EVENT - System shutdown

3. Bidirectional shutdown detection:

   • User closes TrafficLayer (any method) → triggers graceful SUMO close
   • User closes SUMO first → detected via socket error → triggers graceful cleanup

4. Main loop with shutdown flag: Change while(1) to while(!g_shutdownRequested)

Describe alternatives you've considered

• Current approach: Documenting the "correct" shutdown order for users (not ideal)
• Ignoring errors: Suppress error messages but don't actually clean up properly
• Watchdog process: External process to coordinate shutdown (too complex)

Environment (if feature is environment-specific)

• C++ compiler/version: MSVC (Windows-specific console events)
• SUMO version: Using libtraci/libsumo API

Additional context

Implementation Checklist

Files to modify:

• TrafficLayer/TrafficLayer/mainTrafficLayer.cpp

Steps:

1. Add global variables at top of file (before CtrlHandler):

   volatile bool g_shutdownRequested = false;
   TrafficHelper* g_Traffic_c = nullptr;
   SocketHelper* g_Sock_c = nullptr;
   MsgHelper* g_MsgClient_c = nullptr;
   vector<int>* g_actualClientSock = nullptr;
   float* g_simTime = nullptr;
   bool g_initialized = false;

2. Add performCleanup(bool emergencyShutdown) function with try-catch protection

3. Update CtrlHandler() at line ~40:

   • Add cases for CTRL_CLOSE_EVENT, CTRL_LOGOFF_EVENT, CTRL_SHUTDOWN_EVENT
   • Call performCleanup(true) for emergency events
   • Set g_shutdownRequested = true for Ctrl+C/Break

4. In main() around line ~548:

   • Change while(1) to while(!g_shutdownRequested)

5. In main() after connection setup:

   • Initialize global pointers (g_Traffic_c, g_Sock_c, etc.)
   • Set g_initialized = true

6. In recv error handling (line ~610-620):

   • Check for socket errors indicating SUMO disconnection
   • Set g_shutdownRequested = true and break instead of immediate exit(-1)

7. After main loop exits (line ~960):

   • Call performCleanup(false) for normal graceful shutdown
   • Replace existing Traffic_c.close() call

Critical Points

• CTRL_CLOSE_EVENT handling is essential (window X button)
• Windows gives ~5 seconds for close events before force-kill
• Must call libtraci::Simulation::close() BEFORE closing sockets
• Order matters: clients → SUMO → sockets
• All cleanup steps need try-catch for error resilience

Current Code Context

• Current CtrlHandler only handles CTRL_C_EVENT (line 45)
• Main loop is while(1) starting at line 548
• Traffic_c.close() exists at line 960 but sockets close first (wrong order)
• Multiple exit points call Sock_c.socketShutdown() then exit(-1) (scattered throughout)

Related to issue #55 (libsumo/libtraci migration)