Add BigQuery emulator support and fix docker-compose

- Add emulator detection in BigQueryAdapter
  - Use AnonymousCredentials when BIGQUERY_EMULATOR_HOST is set
  - Set API endpoint to emulator using ClientOptions

- Fix docker-compose.yml
  - Remove pre-created dataset (was causing errors)
  - Remove grpc_health_probe healthcheck (not available in image)
  - Use service_started condition for bigquery dependency

Note: BigQuery integration tests currently hang on emulator connection
and need further debugging. Postgres integration tests pass (10/10).

Author: nicosuave

docker-compose.yml

@@ -20,12 +20,7 @@ services:
     platform: linux/amd64
     ports:
       - "9050:9050"
-    command: ["--project=test-project", "--dataset=test_dataset"]
-    healthcheck:
-      test: ["CMD", "grpc_health_probe", "-addr=:9050"]
-      interval: 5s
-      timeout: 5s
-      retries: 5
+    command: ["--project=test-project"]
 
   test:
     build:
@@ -35,7 +30,7 @@ services:
       postgres:
         condition: service_healthy
       bigquery:
-        condition: service_healthy
+        condition: service_started
     environment:
       POSTGRES_TEST: "1"
       POSTGRES_URL: "postgres://test:test@postgres:5432/sidemantic_test"

sidemantic/db/bigquery.py

@@ -77,7 +77,23 @@ def __init__(
                 "Install with: pip install sidemantic[bigquery] or pip install google-cloud-bigquery"
             ) from e
 
-        self.client = bigquery.Client(project=project_id, credentials=credentials, location=location, **kwargs)
+        # Check if using emulator
+        import os
+
+        emulator_host = os.getenv("BIGQUERY_EMULATOR_HOST")
+        if emulator_host:
+            # Use anonymous credentials for emulator
+            from google.auth.credentials import AnonymousCredentials
+            from google.api_core.client_options import ClientOptions
+
+            # Set API endpoint to emulator
+            client_options = ClientOptions(api_endpoint=f"http://{emulator_host}")
+            credentials = AnonymousCredentials()
+            self.client = bigquery.Client(
+                project=project_id, credentials=credentials, location=location, client_options=client_options, **kwargs
+            )
+        else:
+            self.client = bigquery.Client(project=project_id, credentials=credentials, location=location, **kwargs)
         self.project_id = project_id or self.client.project
         self.dataset_id = dataset_id