Fix LocalStack availability for retried PR test jobs

[the-mann]

Problem

When users retry failed jobs in GitHub Actions, the StartLocalStack job may not be rerun if it previously succeeded, but the test jobs that depend on LocalStack will try to connect to a LocalStack instance that no longer exists.

This causes CA bundle tests and other LocalStack-dependent integration tests to fail with connection errors.

Root Cause

• LocalStack runs on EC2 and gets terminated when tests complete
• Retrying individual test jobs doesn't restart the LocalStack dependency
• CA bundle tests fail because they can't reach https://localhost:4566

Solution

Add condition to StartLocalStack job:

if: needs.CheckLabel.outputs.should_run == 'true' || github.run_attempt > 1

How It Works:

• First Run: Only runs if "ready for testing" label is present
• Retry Runs: Always runs to ensure fresh LocalStack instance

Benefits

✅ Retry-Safe: Test jobs always get a running LocalStack instance
✅ Respects Workflow Logic: Still honors label check on first run
✅ Minimal Change: Single line modification
✅ CA Bundle Tests Fixed: LocalStack will always be available for SSL testing

Testing

This fix ensures that when integration tests are retried, they will have access to a fresh LocalStack instance for testing SSL certificate validation and other AWS service interactions.