chore: migrate to gemini-3.1-flash-lite GA model

The Gemini 3.1 Flash Lite preview model is being discontinued on
May 25, 2026. Per Google's GA announcement, the underlying model
architecture is identical and only the model identifier needs to
be updated from `gemini-3.1-flash-lite-preview` to
`gemini-3.1-flash-lite`.

Also relaxes the `_require_gemini_31_preview` guard to accept any
`gemini-3.1-*` identifier (renamed to `_require_gemini_31`), so the
GA name and the still-preview `gemini-3.1-pro-preview` both pass.

tests/test_cli_v3.py

@@ -27,8 +27,8 @@ class CliV3Tests(unittest.TestCase):
             generated_at="2026-03-16T00:00:00+00:00",
             provider_runtime=schema.ProviderRuntime(
                 reasoning_provider="gemini",
-                planner_model="gemini-3.1-flash-lite-preview",
-                rerank_model="gemini-3.1-flash-lite-preview",
+                planner_model="gemini-3.1-flash-lite",
+                rerank_model="gemini-3.1-flash-lite",
             ),
             query_plan=schema.QueryPlan(
                 intent="comparison",

tests/test_evaluator_v3.py

@@ -125,7 +125,7 @@ class EvaluatorV3Tests(unittest.TestCase):
                 topic="test topic",
                 query_type="general",
                 items=[{"key": "a"}],
-                judge_model="gemini-3.1-flash-lite-preview",
+                judge_model="gemini-3.1-flash-lite",
                 gemini_api_key="key",
             )
             self.assertEqual({"a": 3}, cached)
@@ -136,7 +136,7 @@ class EvaluatorV3Tests(unittest.TestCase):
                 topic="test topic",
                 query_type="general",
                 items=[],
-                judge_model="gemini-3.1-flash-lite-preview",
+                judge_model="gemini-3.1-flash-lite",
                 gemini_api_key=None,
             )
             self.assertEqual({}, skipped)

tests/test_render_v3.py

@@ -70,8 +70,8 @@ def sample_report() -> schema.Report:
         generated_at="2026-03-16T00:00:00+00:00",
         provider_runtime=schema.ProviderRuntime(
             reasoning_provider="gemini",
-            planner_model="gemini-3.1-flash-lite-preview",
-            rerank_model="gemini-3.1-flash-lite-preview",
+            planner_model="gemini-3.1-flash-lite",
+            rerank_model="gemini-3.1-flash-lite",
         ),
         query_plan=schema.QueryPlan(
             intent="breaking_news",
@@ -239,8 +239,8 @@ class RenderTopCommentsTests(unittest.TestCase):
             generated_at="2026-03-16T00:00:00+00:00",
             provider_runtime=schema.ProviderRuntime(
                 reasoning_provider="gemini",
-                planner_model="gemini-3.1-flash-lite-preview",
-                rerank_model="gemini-3.1-flash-lite-preview",
+                planner_model="gemini-3.1-flash-lite",
+                rerank_model="gemini-3.1-flash-lite",
             ),
             query_plan=schema.QueryPlan(
                 intent="breaking_news",
@@ -424,8 +424,8 @@ class RenderBestTakesCompactTests(unittest.TestCase):
             generated_at="2026-03-16T00:00:00+00:00",
             provider_runtime=schema.ProviderRuntime(
                 reasoning_provider="gemini",
-                planner_model="gemini-3.1-flash-lite-preview",
-                rerank_model="gemini-3.1-flash-lite-preview",
+                planner_model="gemini-3.1-flash-lite",
+                rerank_model="gemini-3.1-flash-lite",
             ),
             query_plan=schema.QueryPlan(
                 intent="breaking_news",

tests/test_rerank_v3.py

@@ -172,10 +172,10 @@ class RerankV3Tests(unittest.TestCase):
             plan=make_plan(),
             candidates=[first, second],
             provider=provider,
-            model="gemini-3.1-flash-lite-preview",
+            model="gemini-3.1-flash-lite",
             shortlist_size=1,
         )
-        self.assertEqual("gemini-3.1-flash-lite-preview", provider.model)
+        self.assertEqual("gemini-3.1-flash-lite", provider.model)
         self.assertEqual(95.0, first.rerank_score)
         self.assertEqual("high fit", first.explanation)
         # Tail is scored via the fallback (may or may not carry the entity-miss

tests/test_schema_v3.py

@@ -16,8 +16,8 @@ class SchemaV3Tests(unittest.TestCase):
             generated_at="2026-03-16T00:00:00+00:00",
             provider_runtime=schema.ProviderRuntime(
                 reasoning_provider="gemini",
-                planner_model="gemini-3.1-flash-lite-preview",
-                rerank_model="gemini-3.1-flash-lite-preview",
+                planner_model="gemini-3.1-flash-lite",
+                rerank_model="gemini-3.1-flash-lite",
             ),
             query_plan=schema.QueryPlan(
                 intent="breaking_news",