fix: prompt-injection delimiter escape, legacy config coercion, gather resilience, true cross-thread concurrency bound

Addresses items 4-8 from the max-effort review of PR #272 (VectifyAI/PageIndex#272).

- agent.py: wrap_with_doc_context() strips '<'/'>' from doc_name/doc_description
  (untrusted: unsanitized filename / LLM-generated from document content) before
  inserting them into the <docs>...</docs> block, so embedded content can never
  form a literal </docs> that closes the delimiter early and escapes the
  untrusted-data boundary SCOPED_SYSTEM_PROMPT relies on. Deterministic
  per-field transform, doesn't touch the (cacheable) static system prompt.

- ConfigLoader.load() (legacy 0.2.x compat) now routes merged overrides through
  IndexConfig before returning, so a legacy 'no' string gets pydantic's bool
  coercion instead of surviving as a truthy non-empty string — page_index_main's
  bare `if opt.if_add_node_summary:` checks (changed from `== 'yes'` elsewhere
  in this PR) were silently inverting caller intent and firing unwanted billed
  LLM calls.

- verify_toc, process_large_node_recursively, tree_parser,
  generate_summaries_for_structure, generate_summaries_for_structure_md: added
  return_exceptions=True to their asyncio.gather calls (llm_completion/
  llm_acompletion raise RuntimeError on retry exhaustion, added earlier in this
  PR), each with a degrade path matching the pattern already used by sibling
  hardened gathers in the same files. One transient LLM failure no longer
  aborts the whole document's indexing.

- _llm_semaphore is now a true process-wide ceiling (threading.Semaphore,
  shared across every thread/event loop) instead of one asyncio.Semaphore per
  event loop -- concurrently indexing N documents on N threads no longer
  multiplies the effective cap by N. A max_concurrency_scope() override is
  layered as a second, nested, per-loop restriction that can only tighten the
  effective cap within the ceiling, never widen past it.

- set_llm_params() mutated a bare process-wide dict with no per-call isolation,
  unlike max_concurrency which already had ContextVar scoping. Added
  llm_params_scope() (mirrors max_concurrency_scope) + IndexConfig.llm_params,
  wired into build_index() the same way max_concurrency already was, so
  concurrent indexing jobs with different llm kwargs don't leak into each
  other.

Adds regression tests for all five. Full suite: 221 passed, 2 skipped (one
pre-existing, unrelated flaky cloud-streaming test intermittently fails on
rerun; confirmed independent of this change).

Claude-Session: https://claude.ai/code/session_01Kx5DgKbhK1N8autqXH8SmS

pageindex/agent.py

@@ -46,20 +46,29 @@ Answer based only on tool output. Be concise.
 """
 
 
+def _defang_delimiters(text: str) -> str:
+    """Strip '<'/'>' so untrusted text can never form a literal <docs>/</docs>
+    (or any other tag-shaped string) that would prematurely close the
+    wrap_with_doc_context() delimiter and escape the untrusted-data boundary."""
+    return text.replace("<", "").replace(">", "")
+
+
 def wrap_with_doc_context(docs: list[dict], question: str) -> str:
     """Prepend a doc-context block to the user question for scoped queries.
 
     Document fields (especially doc_description, which is LLM-generated at
     index time) are untrusted text that may contain adversarial instructions.
     We wrap them in a <docs>...</docs> delimiter and tell the agent in the
-    system prompt to treat the block as data only.
+    system prompt to treat the block as data only. '<'/'>' are stripped from
+    the untrusted fields first so embedded content can never form a literal
+    </docs> (or any other tag) that closes the delimiter early.
     """
     lines = []
     for d in docs:
-        line = f"- {d['doc_id']}: {d.get('doc_name', '')}"
+        line = f"- {d['doc_id']}: {_defang_delimiters(d.get('doc_name', ''))}"
         desc = d.get("doc_description") or ""
         if desc:
-            line += f" — {desc}"
+            line += f" — {_defang_delimiters(desc)}"
         lines.append(line)
     label = "document" if len(docs) == 1 else "documents"
     return (