# Change: Add Concurrency Handling and Reliability Improvements

## Why
Multiple users editing the same task simultaneously can lead to lost updates (last-write-wins). Trigger execution failures are not retried, causing permanent failures on transient errors. Soft-delete restore operation does not cascade to child tasks, leading to data inconsistency.

## What Changes
- Implement optimistic locking using version field for task updates
- Add retry mechanism with exponential backoff for trigger execution
- Implement cascade restore for soft-deleted tasks and their children

## Impact
- Affected specs: task-management, automation
- Affected code:
  - `backend/app/models/task.py` - Add version field
  - `backend/app/api/tasks/router.py` - Optimistic locking logic
  - `backend/app/services/trigger_scheduler.py` - Retry mechanism
  - `backend/app/api/tasks/router.py` - Cascade restore