chore(aws): Increase healthcheck timeout to 15 mins

[Antonio-RiveroMartnez]

Problem

Showtime deployments are failing health checks even though the container is running correctly. The health check times out before the container becomes ready.

Root cause: Superset's Parquet modernization (PR #36538) merged in January 2026 changed how example data is loaded. Instead of using a pre-built DuckDB file, 23 datasets (~400k rows) are now loaded at runtime from Parquet files. Making the whole startup process take 10+ minutes, but showtime's health check timeout was only 10 minutes.

Timeline of a failed deployment
Time Event
T+0 Container starts, ECS marks service "stable"
T+0 Superset begins loading examples (blocking)
T+10min Showtime health check times out (20 attempts exhausted)
T+10+min Examples finish loading, gunicorn starts
T+10+min Container now healthy (too late)

This PR Increases max_attempts from 20 to 30 (10 min → 15 min timeout) and adds some unit tests.

Another approach we could take but it implies more complexity and connecting new aws clients, permissions etc:

1. ECS tasks write logs to CloudWatch (typically /ecs/superset-ci/ or similar)
2. Watch for gunicorn startup message like "Listening at: http://0.0.0.0:8088" or "Booting
   worker with pid"
3. Only then start HTTP health checks

For example This PR: apache/superset#37694 is marked as fail but the env is actually up and working: http://34.219.183.81:8080/

  PR #37694 (SHA: b427b82)                                                                        
      ↓                                                                                           
  ECS Service: pr-37694-b427b82-service                                                           
      ↓                                                                                           
  ECS Task: de146f58caf14cafbae0b62fa8f8855c                                                      
      ↓                                                                                           
  Network Interface: eni-035b411082b95b518                                                        
      ↓                                                                                           
  Public IP: 34.219.183.81                                                                        
      ↓
  Health: OK ✅