Provide mechanism to clean up any failed deployments of resources

Describe the bug

make db-migrate can fail where cosmos contains resources that correspond to workspaces that have failed to deploy. api logs:

2023-02-10T10:09:39.761582740Z INFO:root:PR 1717 - Shared services
2023-02-10T10:09:39.824579237Z INFO:root:PR 1726 - Authentication needs to be in properties so we can update them
2023-02-10T10:09:39.832641465Z ERROR:root:Failed to migrate database
2023-02-10T10:09:39.832683565Z Traceback (most recent call last):
2023-02-10T10:09:39.832688665Z File "/api/./api/routes/migrations.py", line 51, in migrate_database
2023-02-10T10:09:39.832692265Z migration_status = "Executed" if await workspace_migration.moveAuthInformationToProperties() else "Skipped"
2023-02-10T10:09:39.832695665Z File "/api/./db/migrations/workspaces.py", line 31, in moveAuthInformationToProperties
2023-02-10T10:09:39.832698965Z item["properties"]["scope_id"] = f"api://{item['properties']['client_id']}"
2023-02-10T10:09:39.832702365Z KeyError: 'client_id'
2023-02-10T10:09:39.835888016Z {'paths': ['/id'], 'kind': 'Hash', 'version': 2}
2023-02-10T10:09:39.835919516Z {'paths': ['/id'], 'kind': 'Hash', 'version': 2}
2023-02-10T10:09:39.835924616Z {'paths': ['/resourceId'], 'kind': 'Hash', 'version': 2}
2023-02-10T10:09:39.835927617Z {'paths': ['/id'], 'kind': 'Hash', 'version': 2}
2023-02-10T10:09:39.835930517Z {'paths': ['/id'], 'kind': 'Hash', 'version': 2}
2023-02-10T10:09:39.835933417Z {'paths': ['/id'], 'kind': 'Hash', 'version': 2}
2023-02-10T10:09:39.835936417Z {'paths': ['/id'], 'kind': 'Hash', 'version': 2}
2023-02-10T10:09:39.835939317Z INFO: x.x.x.x:x - "POST /api/migrations HTTP/1.1" 400 Bad Request

Deleting the workspaces without a client_id field from cosmos fixes the issue

Steps to reproduce

1. Deploy a workspace that does not succeed, due to an unrelated issue
2. Run make db-migrate

Start with api/routes/migrations.py and db/migrations/workspaces.py, then reproduce the failure with make db-migrate using Cosmos resources from a failed workspace. Done means failed workspace resources without a client_id are cleaned up or safely handled so the migration completes without the reported KeyError.