feat: Complete multi-client runner implementation with pubsub-sub-bench integration

Major improvements to multi-client execution:

🚀 Multi-Client Runner Enhancements:
- Remove unnecessary threading (containers already detached)
- Add proper error handling with container logs output
- Fix permission issues with Docker user parameter
- Implement JSON result aggregation from multiple containers
- Add fail-fast behavior on container failures

🔧 Container Management:
- Add configurable --container-timeout-buffer flag (default 60s)
- Fix Docker container user permissions for file writing
- Improve container lifecycle management and cleanup

📊 Results Processing:
- Fix filename mismatch between multi-client JSON output and processing
- Add proper metric extraction for multi-client scenarios
- Fix missing final summary table in multi-client output
- Ensure both individual and aggregate results tables display

🧪 Testing:
- Add comprehensive tests for new functionality
- Test container timeout buffer configuration
- Verify multi-client execution with memtier + pubsub-sub-bench

✅ Full Feature Parity:
Multi-client runner now has complete feature parity with single-client:
- Real performance metrics extraction (114K+ ops/sec)
- Complete error handling and logging
- Proper results aggregation and display
- Support for memtier_benchmark + pubsub-sub-bench simultaneously

All tests passing with both single and multi-client configurations.

Author: fcostaoliveira

redis_benchmarks_specification/__runner__/args.py

@@ -208,6 +208,12 @@ def create_client_runner_args(project_name):
         type=int,
         help="override memtier number of runs for each benchmark. By default will run once each test",
     )
+    parser.add_argument(
+        "--container-timeout-buffer",
+        default=60,
+        type=int,
+        help="Buffer time in seconds to add to test-time for container timeout. Default is 60 seconds.",
+    )
     parser.add_argument(
         "--cluster-mode",
         default=False,