Resolve the Solar throttle gap from env with a 32-wide fallback 🟩

Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>

applications/modelling_e2e/handler.py

@@ -296,7 +296,13 @@ def _solar_min_request_interval_seconds() -> float:
     the queue's ``maximum_concurrency`` (0.8 × 10 QPS ÷ N) so the up-to-32-wide
     fleet stays under the hard 600 QPM Solar ceiling. Falls back to the 32-wide
     default when unset or unparseable."""
-    raise NotImplementedError
+    raw = os.environ.get("SOLAR_MIN_REQUEST_INTERVAL_SECONDS")
+    if raw is None:
+        return _DEFAULT_SOLAR_MIN_REQUEST_INTERVAL_SECONDS
+    try:
+        return float(raw)
+    except ValueError:
+        return _DEFAULT_SOLAR_MIN_REQUEST_INTERVAL_SECONDS
 
 
 def _solar_insights_for(
@@ -417,7 +423,10 @@ def handler(body: dict[str, Any], context: Any, orchestrator: TaskOrchestrator,
     engine = _get_engine()
     epc_client = EpcClientService(os.environ["OPEN_EPC_API_TOKEN"])
     geospatial = GeospatialS3Repository(_s3_parquet_reader())
-    solar_client = GoogleSolarApiClient(os.environ["GOOGLE_SOLAR_API_KEY"])
+    solar_client = GoogleSolarApiClient(
+        os.environ["GOOGLE_SOLAR_API_KEY"],
+        min_request_interval_seconds=_solar_min_request_interval_seconds(),
+    )
 
     with engine.connect() as conn:
         property_rows = conn.execute(