feat: add running operations tracking to prevent unsafe termination

[ParidelPooya]

Running Operations Tracking

What We Built:

1. hasRunningOperations() method - Clean API to check if critical operations are running
2. Internal tracking system - addRunningOperation() / removeRunningOperation() with Set-based storage
3. Automatic cleanup - try/finally blocks ensure operations are removed even on errors

Where We Added Tracking (Only 2 Operations):

✅ ctx.step:
• Executes user code
• Critical for workflow state - if interrupted, workflow state becomes inconsistent

✅ ctx.waitForCondition:
• Executes user condition checks
• Critical for workflow logic - if interrupted, condition evaluation becomes unreliable

Why NOT Other Operations:

❌ ctx.runInChildContext - Safe to interrupt, just context management
❌ ctx.wait - Simple timer, no user code execution
❌ ctx.createCallback - Just creates callback config, no execution
❌ ctx.waitForCallback - Uses ctx.step internally, so already tracked automatically
❌ ctx.map/parallel/executeConcurrently - Use ctx.runInChildContext internally, which is safe to interrupt

Key Insight:

We only needed to add tracking to the 2 fundamental operations that execute user code:
• ctx.step - Direct user code execution
• ctx.waitForCondition - User condition evaluation

All other operations either:
• Are safe to interrupt (context/config operations)
• Delegate to tracked operations (so get protection automatically)

The Result:

// Safe termination check
if (durableContext.hasRunningOperations()) {
  return; // Critical user code is running - wait
}
terminate(); // Safe to terminate

Surgical precision - Only the 2 operations that actually execute user code are tracked! 🎯