add-node workflow: fix recovery and wal restore issues, improve clean up logic

[mmols]

• Resolved an issue where recovery.signal would remain after the add-node process completed by setting --type=standby in the pgBackRest restore
• Fixed an issue where cipher-type would not be set in the restore_command due to how pgBackRest generates the command, resulting in the inability to restore WAL from the repository (if needed)
• Refactored add-node process to cleanly refer to source and target node
• Simplified pgbackrest reconfguration logic for the source / target nodes
• Added a backrest cleanup-replica command that ensures the newly added node does not have lingering configuration from the add node process
• Ensured the restore_command is unset on the target node after the add node process completes
• Fixed an issue where the pgBackRest yaml was rewritten to all nodes during add-node. It will now only be written to the target node during reconfiguration if pgBackRest should be configured on the target node
• Removed logic that set S3 environment variables on the target node. This is documented as a prerequisite and will be handled by the user
• Removed unused functions in backrest.py
• Fixed an issue where repo-type S3 would not have it's repository configured
• Fixed an issue where cipher-type=none was not respected
• Fixed an issue where ssl_cert_file, ssl_key_file, and log_directory would not be properly configured on the target node
• Removed code that set shared_preload_libraries on the target node - this is already set during the restore
• Fixed an issue where add-node would use public ips for replication connections, even if private ips were provided
• Add polling logic to verify postgres comes out of recovery during add node

Resolves PLAT-53, PLAT-34, PLAT-45

[moizpgedge]

Ready to be merged