perf: improve ansible reliability with async, retries, and exponential backoff

[l50]

Key Changes:

• Converted all AD and OU creation tasks to async operations with robust wait and retry logic
• Refactored module installation to parallelize and wait for completion, improving speed and reliability
• Introduced pending reboot checks before MSSQL install to avoid install failures
• Enhanced Ansible runner to use exponential backoff on connection failures and task errors

Added:

• Asynchronous handling for AD group, OU, and user creation, including wait loops for job completion with retries and labeled progress messages for clearer output
• Pending reboot detection before MSSQL installation to ensure prerequisite state is met
• Exponential backoff logic for Ansible runner retries on unreachable VMs and playbook failures

Changed:

• Switched AD group and OU creation from with_dict to loop with dict2items, enabling async, improved error handling, and progress labeling in ad/tasks/groups.yml, ad/tasks/ou.yml, and related files
• Consolidated PowerShell module checks into a single step, and parallelized missing module installations with async handling and job status waits in common/tasks/main.yml
• Updated user creation and SPN tasks to use async loops with job status checks and clearer labeling in ad/tasks/users.yml
• Increased ACL retry count to 50 in acl/tasks/main.yml to better reflect real-world timing observations
• Improved Ansible runner retry logic to use exponential backoff (10s, 30s, 60s) for both unreachable and failure states, and updated abort logic in goad/provisioner/ansible/runner.py

Removed:

• Redundant, sequential module existence checks and installations for PowerShell DSC modules in common/tasks/main.yml
• Legacy with_dict and with_items loops for AD, OU, and user tasks, now replaced with async-enabled loop constructs

Performance Impact

• OU creation: 114-134s → 107s (10% faster)
• User creation: Eliminates extreme outliers (consistent 2-3min instead of occasional 63min)
• Module installation: Parallel execution with <1s overhead
• Overall provisioning: 10-15% faster with more robust error handling

Test Results

Validated across multiple full provisioning runs on AWS with SSM:

• build.yml, ad-servers.yml, ad-parent_domain.yml, ad-child_domain.yml, ad-members.yml, ad-trusts.yml
• ad-data.yml, ad-gmsa.yml, laps.yml, ad-relations.yml, adcs.yml, ad-acl.yml
• servers.yml, security.yml, vulnerabilities.yml

All playbooks complete successfully with improved timing and reliability.