fix: correct remaining API mismatches in pipeline, vector_store, and normalize docs

- pipeline.md: ParallelismManager pool_type="thread"/"process" → use_processes=False/True;
  execute_parallel() returns List[ParallelExecutionResult] not aggregate object
- vector_store.md: remove MetadataStore.add_field() (method is on MetadataSchema, not
  MetadataStore); fix tip to reference MetadataStore.update_metadata() not VectorStore
- normalize.md: Pipeline() orchestrator misuse → PipelineBuilder + ExecutionEngine pattern

docs/reference/normalize.md

@@ -360,21 +360,27 @@ print(f"Invalid: {result.error_count}")
 ## Pipeline Integration
 
 ```python
-from semantica.pipeline import Pipeline
+from semantica.pipeline import PipelineBuilder, ExecutionEngine
 from semantica.ingest import FileIngestor
 from semantica.normalize import TextNormalizer
 from semantica.semantic_extract import NERExtractor
 from semantica.llms import Groq
 import os
 
-llm = Groq(model="llama-3.3-70b-versatile", api_key=os.getenv("GROQ_API_KEY"))
+llm       = Groq(model="llama-3.3-70b-versatile", api_key=os.getenv("GROQ_API_KEY"))
+ingestor  = FileIngestor()
+normalizer = TextNormalizer(strip_html=True, normalize_unicode=True)
+extractor = NERExtractor(method="llm", llm_provider=llm)
 
-pipeline = Pipeline()
-pipeline.add_step("ingest",    FileIngestor())
-pipeline.add_step("normalize", TextNormalizer(strip_html=True, normalize_unicode=True))
-pipeline.add_step("extract",   NERExtractor(method="llm", llm_provider=llm))
+builder = PipelineBuilder()
+builder.add_step("ingest",    "file_ingest",    handler=ingestor.ingest)
+builder.add_step("normalize", "text_normalize", handler=normalizer.normalize)
+builder.add_step("extract",   "ner_extract",    handler=extractor.extract)
+builder.connect_steps("ingest",    "normalize")
+builder.connect_steps("normalize", "extract")
 
-result = pipeline.run("data/documents/")
+pipeline = builder.build("normalize_pipeline")
+result   = ExecutionEngine().execute_pipeline(pipeline, data="data/documents/")
 ```
 
 ## Tips and Common Pitfalls

docs/reference/pipeline.md

@@ -405,25 +405,29 @@ Checks performed:
     ```python
     from semantica.pipeline import ParallelismManager
 
-    manager = ParallelismManager(max_workers=8, pool_type="thread")
+    # use_processes=False (default) → thread pool for I/O-bound tasks
+    manager = ParallelismManager(max_workers=8, use_processes=False)
 
-    tasks  = [{"fn": ner.extract, "args": [text]} for text in texts]
-    result = manager.execute_parallel(tasks, timeout=60)
+    tasks   = [{"fn": ner.extract, "args": [text]} for text in texts]
+    results = manager.execute_parallel(tasks, timeout=60)
+    # returns List[ParallelExecutionResult]
 
-    print(f"Successful: {result.success_count}, Failed: {result.failure_count}")
+    successes = [r for r in results if r.success]
+    failures  = [r for r in results if not r.success]
     ```
 
-    Use thread pools for **I/O-bound** steps: web fetching, database queries, API calls. Threads share memory and context-switch cheaply between waiting operations.
+    Use thread pools for **I/O-bound** steps: web fetching, database queries, API calls.
   </Tab>
   <Tab title="Process pool (CPU-bound)">
     ```python
-    manager = ParallelismManager(max_workers=4, pool_type="process")
+    # use_processes=True → process pool, bypasses Python GIL
+    manager = ParallelismManager(max_workers=4, use_processes=True)
 
-    tasks  = [{"fn": embedder.embed, "args": [chunk]} for chunk in chunks]
-    result = manager.execute_parallel(tasks, timeout=120)
+    tasks   = [{"fn": embedder.generate_embeddings, "args": [chunk]} for chunk in chunks]
+    results = manager.execute_parallel(tasks, timeout=120)
     ```
 
-    Use process pools for **CPU-bound** steps: embedding computation, OCR, large NER batches. Processes bypass Python's GIL for true multi-core parallelism.
+    Use process pools for **CPU-bound** steps: embedding, OCR, large NER batches.
   </Tab>
 </Tabs>
 

docs/reference/vector_store.md

@@ -317,12 +317,6 @@ from semantica.vector_store import MetadataStore
 
 meta_store = MetadataStore()
 
-# Define schema fields
-meta_store.add_field("author",   str,   required=True)
-meta_store.add_field("year",     int,   required=True)
-meta_store.add_field("category", str)
-meta_store.add_field("score",    float, default=0.0)
-
 # Store and retrieve metadata
 meta_store.store_metadata("doc1", {"author": "Alice", "year": 2024, "category": "research"})
 meta_store.store_metadata("doc2", {"author": "Bob",   "year": 2023, "category": "review"})
@@ -423,7 +417,7 @@ store = VectorStore(backend="faiss", dimension=768, metric="cosine")
 </Warning>
 
 <Tip>
-  **Update metadata without re-embedding.** `store.update_metadata(id, {...})` changes attached fields (status, tags, review date) without re-running the embedding model. Use this for state changes that don't affect semantic content.
+  **Update metadata without re-embedding.** `MetadataStore.update_metadata(id, {...})` changes attached fields (status, tags, review date) without re-running the embedding model. Use this for state changes that don't affect semantic content.
 </Tip>
 
 <CardGroup cols={2}>