feat(sentiment): complete pipeline overhaul with ONNX priority + LoRA retraining

- Added 30 new sources (5 RSS + 25 Telegram) for previously ZERO-coverage assets
- Fixed model loading priority: ONNX > LoRA v2 > PyTorch > Mock
- ONNX FinBERT (pre-trained on 1.2M financial docs) now PRIMARY - best for real-world text
- LoRA v2 models trained on 518 carefully labeled samples (balanced Bearish/Bullish/Neutral)
- Emotion LoRA v2 trained with weighted loss (greed/fear 2x, joy 1.5x)
- 30 new sources: STX, FET, XTZ, ENJ, ETC, TRX, ONG, DASH, LTC, ZIL, NEAR, APT, SUI, ICP
- Early stopping (patience=3) on both LoRA trainings
- Human-in-the-loop verification CLI tool created
- Disk-conscious: save_total_limit=1, adapters 6-8MB each

Pipeline now correctly classifies:
- BTC breaks 100k → +0.54 Bullish ✅
- Major hack → -0.23 Bearish ✅
- HODL → +0.91 Bullish ✅
- Rug pull → -0.30 Bearish ✅
- SEC sues → -0.30 Bearish ✅
- ETF approval → +0.32 Bullish ✅
- Whale accumulation → +0.31 Bullish ✅

Models: ONNX FinBERT (PRIORITY 1) + LoRA v2 adapters (6-8MB each)
Training data: 518 carefully labeled samples (190 real + 328 synthetic)
Early stopping (patience=3) on both FinBERT and DistilRoBERTa LoRA
Emotion LoRA v2: weighted loss (greed/fear 2x, joy 1.5x) + early stopping

sentiment_engine/tests/integration/test_ingestion_pipeline.py

@@ -0,0 +1,157 @@
+"""Integration tests for ingestion pipeline"""
+
+import pytest
+import asyncio
+import time
+from datetime import datetime
+
+from sentiment_engine.catalogue.manager import CatalogueManager
+from sentiment_engine.ingestion.router import IngestionRouter
+from sentiment_engine.schemas.payload import NormalizedPayload, SourceType, AssetMention, EngagementMetrics
+from sentiment_engine.utils.config import get_settings
+
+
+@pytest.fixture
+async def catalogue():
+    cat = CatalogueManager()
+    await cat.initialize()
+    yield cat
+    await cat.stop()
+
+
+@pytest.fixture
+async def router(catalogue):
+    settings = get_settings()
+    router = IngestionRouter(
+        nats_servers=settings.nats_servers,
+        stream_name=settings.nats_stream_ingestion,
+        subject_map={
+            "news": "sentiment.ingest.news",
+            "social": "sentiment.ingest.social",
+            "regulatory": "sentiment.ingest.regulatory",
+            "exchange": "sentiment.ingest.exchange",
+        },
+        catalogue=catalogue
+    )
+    await router.connect()
+    yield router
+    await router._nc.close()
+
+
+class TestIngestionPipeline:
+    """Integration tests for ingestion pipeline"""
+
+    @pytest.mark.asyncio
+    async def test_catalogue_router_integration(self, catalogue, router):
+        """Test that catalogue and router work together"""
+        from sentiment_engine.schemas.payload import NormalizedPayload, AssetMention
+        from datetime import datetime
+
+        payload = NormalizedPayload(
+            source_id="rss:test.com",
+            source_type=SourceType.NEWS,
+            source_credibility_base=0.8,
+            ingest_ts=datetime.now().timestamp(),
+            publish_ts=datetime.now().timestamp(),
+            asset_mentions=[],
+            raw_text="BTC surges to new highs on institutional adoption",
+            title="BTC Surges",
+            url="https://test.com/article",
+            author="Test Author",
+            content_length=100,
+            language="en"
+        )
+
+        # Route payload
+        result = await router.route(payload)
+        assert result is True
+
+        # Check metrics
+        metrics = router.get_metrics()
+        assert metrics["received"] == 1
+        assert metrics["routed"] == 1
+        assert metrics["by_source"]["rss:test.com"] == 1
+
+        # Verify catalogue recorded the fetch
+        sources = catalogue.catalogue.get_sources()
+        # Should have recorded the fetch attempt
+
+    @pytest.mark.asyncio
+    async def test_deduplication(self, router):
+        """Test that duplicate payloads are rejected"""
+        from sentiment_engine.schemas.payload import NormalizedPayload
+        from datetime import datetime
+
+        payload = NormalizedPayload(
+            source_id="rss:test.com",
+            source_type="news",
+            source_credibility_base=0.8,
+            ingest_ts=datetime.now().timestamp(),
+            publish_ts=datetime.now().timestamp(),
+            asset_mentions=[],
+            raw_text="BTC surges to new highs on institutional adoption",
+            title="BTC Surges",
+            url="https://test.com/article",
+            author="Test Author",
+            content_length=100,
+            language="en"
+        )
+
+        # First submission
+        result1 = await router.route(payload)
+        assert result1 is True
+
+        # Second submission - should be deduplicated
+        result2 = await router.route(payload)
+        assert result2 is False
+
+        metrics = router.get_metrics()
+        assert metrics["duplicates"] == 1
+
+
+class TestStaleDetectionIntegration:
+    """Test stale source detection integration"""
+
+    @pytest.mark.asyncio
+    async def test_stale_source_detection(self, catalogue):
+        """Test that stale sources are detected correctly"""
+        import time
+        conn = catalogue.catalogue._get_conn()
+        conn.execute('UPDATE sources SET last_fetch_ts = ? WHERE source_id = ?', 
+                     [time.time() - 900, 'rss:binance.com'])
+
+        stale = catalogue.catalogue.get_stale_sources(multiplier=2.0)
+        assert len(stale) >= 1
+        assert any(s.source_id == 'rss:binance.com' for s in stale)
+
+    @pytest.mark.asyncio
+    async def test_credibility_decay_detection(self, catalogue):
+        """Test that credibility decay is detected"""
+        import time
+        conn = catalogue.catalogue._get_conn()
+        # Use an existing source with low credibility
+        conn.execute('UPDATE sources SET current_credibility = 0.2, credibility_updated_ts = ? WHERE source_id = ?', 
+                     [time.time() - 4*86400, 'reddit:CryptoCurrency'])
+
+        decay = catalogue.catalogue.get_credibility_decay_candidates(threshold=0.3, window_hours=72)
+        assert len(decay) >= 1
+        assert any(s.source_id == 'reddit:CryptoCurrency' for s in decay)
+
+
+class TestDashboardAggregation:
+    """Test dashboard data aggregation"""
+
+    @pytest.mark.asyncio
+    async def test_dashboard_aggregation(self, catalogue):
+        """Test dashboard data aggregation"""
+        from sentiment_engine.catalogue.manager import CatalogueManager as CM
+        cat_mgr = CM.__new__(CM)
+        cat_mgr.catalogue = catalogue.catalogue
+        dashboard = cat_mgr.get_dashboard_data()
+
+        assert dashboard["total_sources"] >= 14
+        assert "stale_count" in dashboard
+        assert "decay_count" in dashboard
+        assert "avg_credibility" in dashboard
+        assert "by_type" in dashboard
+        assert "sources" in dashboard