fix(llm): default generate_completion num_ctx to Config.OLLAMA_NUM_CTX

Non-streaming generate_completion was the only LLM entry point that
didn't default num_ctx — stream_chat and stream_chat_with_tools both
fall back to Config.OLLAMA_NUM_CTX (16384). When a caller omitted the
argument, Ollama silently used the model's default window (~4k on
qwen3) and truncated the prompt.

That footgun was masked by fallback paths in the research pipeline:
_generate_outline's prompt carries ~12 sources × 2000 chars (~6k
tokens) of source material plus a system prompt, so the prompt got
chopped, the model never saw the sources, JSON parsing failed twice,
and run_research_pipeline dropped into the single-note "monolith"
fallback (research.py:251). The user reported chat 215 producing such
a monolith note for a multi-source research topic.

Two-layer fix:
- Default num_ctx to Config.OLLAMA_NUM_CTX inside generate_completion,
  matching the streaming entry points. Any current or future caller
  that forgets the argument stops silently losing input.
- Pin num_ctx=16384 explicitly in _generate_outline and
  _generate_executive_summary with comments pointing at the failure
  mode, so a refactor of the generate_completion default can't
  silently regress the research pipeline.

Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>

src/fabledassistant/services/llm.py

@@ -317,9 +317,17 @@ async def generate_completion(
     num_ctx overrides the model's context window for this call only.
     """
     last_exc: Exception | None = None
-    options: dict = {"num_predict": max_tokens}
-    if num_ctx is not None:
-        options["num_ctx"] = num_ctx
+    # Default num_ctx to Config.OLLAMA_NUM_CTX (matching stream_chat /
+    # stream_chat_with_tools). Without this, Ollama silently uses the model's
+    # default window (~4k on qwen3) and truncates anything longer. That is
+    # how the research pipeline's outline step kept falling back to a single
+    # monolith note: its 12-source prompt is ~6k tokens and was being chopped
+    # before the model ever saw it. Non-streaming callers must not inherit
+    # that footgun — if you truly want the model default, pass num_ctx=0.
+    options: dict = {
+        "num_predict": max_tokens,
+        "num_ctx": num_ctx if num_ctx is not None else Config.OLLAMA_NUM_CTX,
+    }
     for attempt in range(3):
         if attempt > 0:
             delay = 3.0 * attempt