Updated tech specs for:

- Tool candidates on Analysis events
- Token counts — per-event on provenance, aggregated on service
  responses, --show-usage on all CLIs
- Termination reason on terminal events
- Tool error/retry events with Error mixin
- Sub-agent linkage (was already correct)
- Vocabulary publication (OWL ontology)
- Step counter on iteration events
- Pattern decision entity in DAG
- Latency breakdown (LLM + tool duration)
- Envelope field naming unified (message_type)
- LLM parse errors handled gracefully

docs/tech-specs/python-api-refactor.md

@@ -704,17 +704,17 @@ class StreamingChunk:
 @dataclasses.dataclass
 class AgentThought(StreamingChunk):
     """Agent reasoning chunk"""
-    chunk_type: str = "thought"
+    message_type: str = "thought"
 
 @dataclasses.dataclass
 class AgentObservation(StreamingChunk):
     """Agent tool observation chunk"""
-    chunk_type: str = "observation"
+    message_type: str = "observation"
 
 @dataclasses.dataclass
 class AgentAnswer(StreamingChunk):
     """Agent final answer chunk"""
-    chunk_type: str = "final-answer"
+    message_type: str = "final-answer"
     end_of_dialog: bool = False
 
 @dataclasses.dataclass