feat(oss): port v3 pipeline with hybrid search, entity extraction, and additive scoring

Replace the 2-LLM-call add pipeline with a single-pass additive extraction
using ADDITIVE_EXTRACTION_PROMPT. Memories now accumulate (ADD-only) instead
of being updated/deleted during extraction.

Search pipeline upgraded to hybrid scoring combining three signals:
- Semantic similarity (vector search)
- BM25 keyword matching (native per vector store, 15 stores supported)
- Entity boost (spaCy NER with entity collection linking)

Combined score = (semantic + bm25 + entity_boost) / max_possible, where
max_possible adapts based on which signals are active.

Key changes:
- Add spaCy-based lemmatization for BM25 keyword search
- Add spaCy-based entity extraction (PROPER, QUOTED, COMPOUND, NOUN types)
- Add entity store as second vector collection ({collection}_entities)
- Add native keyword_search() to 15 vector store adapters
- Add batch embedding support (embed_batch) for OpenAI and Azure OpenAI
- Add message persistence in SQLite (rolling window of 10 per scope)
- Add additive scoring with adaptive normalization
- Add observation_date parameter to add()
- Add custom_instructions config field
- Default search threshold changed to 0.1

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

tests/utils/test_entity_extraction.py

@@ -0,0 +1,102 @@
+import pytest
+
+
+@pytest.fixture(autouse=True)
+def _ensure_spacy():
+    """Skip tests if spaCy model is not available."""
+    try:
+        import spacy
+        spacy.load("en_core_web_sm")
+    except Exception:
+        pytest.skip("spaCy en_core_web_sm model not available")
+
+
+class TestExtractEntities:
+    def test_proper_nouns(self):
+        from mem0.utils.entity_extraction import extract_entities
+
+        entities = extract_entities("John Smith works at Google on machine learning projects")
+        entity_texts = [e[1] for e in entities]
+        entity_types = [e[0] for e in entities]
+        # Should extract proper nouns
+        found_proper = any("John" in t or "Google" in t for t in entity_texts)
+        assert found_proper, f"Expected proper nouns, got {entities}"
+
+    def test_quoted_text(self):
+        from mem0.utils.entity_extraction import extract_entities
+
+        entities = extract_entities('She is reading "The Great Gatsby" this week')
+        entity_texts = [e[1] for e in entities]
+        assert any("Great Gatsby" in t for t in entity_texts), f"Expected quoted text, got {entities}"
+
+    def test_compound_nouns(self):
+        from mem0.utils.entity_extraction import extract_entities
+
+        entities = extract_entities("The machine learning engineer built a neural network")
+        entity_texts = [e[1].lower() for e in entities]
+        has_compound = any("machine" in t and "learning" in t for t in entity_texts) or \
+                       any("neural" in t and "network" in t for t in entity_texts)
+        assert has_compound, f"Expected compound nouns, got {entities}"
+
+    def test_empty_string(self):
+        from mem0.utils.entity_extraction import extract_entities
+
+        entities = extract_entities("")
+        assert entities == []
+
+    def test_no_entities(self):
+        from mem0.utils.entity_extraction import extract_entities
+
+        entities = extract_entities("I like things and stuff")
+        # Generic words should be filtered out
+        entity_texts = [e[1].lower() for e in entities]
+        assert "things" not in entity_texts
+        assert "stuff" not in entity_texts
+
+    def test_deduplication(self):
+        from mem0.utils.entity_extraction import extract_entities
+
+        entities = extract_entities("Google is great. I love working at Google.")
+        google_count = sum(1 for _, t in entities if "Google" in t)
+        assert google_count <= 1, f"Expected dedup, got {entities}"
+
+    def test_returns_tuples(self):
+        from mem0.utils.entity_extraction import extract_entities
+
+        entities = extract_entities("John Smith lives in New York City")
+        for entity in entities:
+            assert isinstance(entity, tuple)
+            assert len(entity) == 2
+            assert entity[0] in ("PROPER", "QUOTED", "COMPOUND", "NOUN")
+            assert isinstance(entity[1], str)
+
+
+class TestExtractEntitiesBatch:
+    def test_batch_processing(self):
+        from mem0.utils.entity_extraction import extract_entities_batch
+
+        texts = [
+            "John works at Google",
+            "Mary lives in Paris",
+            "The cat sat on the mat",
+        ]
+        results = extract_entities_batch(texts)
+        assert len(results) == 3
+        assert isinstance(results[0], list)
+        assert isinstance(results[1], list)
+        assert isinstance(results[2], list)
+
+    def test_empty_input(self):
+        from mem0.utils.entity_extraction import extract_entities_batch
+
+        assert extract_entities_batch([]) == []
+
+    def test_consistency_with_single(self):
+        from mem0.utils.entity_extraction import extract_entities, extract_entities_batch
+
+        text = "John Smith works at Google headquarters"
+        single = extract_entities(text)
+        batch = extract_entities_batch([text])
+        assert len(batch) == 1
+        # Both should extract the same entities
+        assert set(t for _, t in single) == set(t for _, t in batch[0])