qa/corpus_query: wikitext-strip when policy.base_version is set (6d)

Closes Phase 1 step 6 of #53. When the caller passes
policy={"base_version": "wikitext-base-v1"} (or any truthy value),
run_query now invokes arborist.wikitext.to_base() on each chunk
span BEFORE truncating to per_doc_budget, so a 24 kB raw-template
span gets compressed first and the LLM/verifier see the same
prose. Same shape legacy query() applies at three sites.

policy=None / policy without base_version: no-op, preserves the
byte-identity gate. mwparserfromhell missing (the optional dep):
import fails, no-op — already the legacy convention.

policy["base_version"] folds into governance_policy_hash via the
caller's policy dict — opting in DOES rotate cache_key for that
policy partition, which is the right semantics (different prose →
different cache).

264 tests still pass. Phase 1 substantively complete; Phase 2
(providence_query.py + legacy query() collapse) starts next.

arborist/qa/corpus_query.py

@@ -303,6 +303,19 @@ def run_query(
         per_hit_budgets = [
             max(1000, max_context_chars // max(1, len(hits)))
         ] * len(hits)
+    # Wikitext-strip (step 6d): when policy["base_version"] is set
+    # AND the wikitext module imports, run to_base() on each span
+    # BEFORE truncating to per_doc_budget so a 24kB raw-template span
+    # gets compressed first. Same shape legacy query() uses at three
+    # sites; folds into governance_policy_hash via policy["base_version"]
+    # — caller's responsibility to pin the version field if they want
+    # cache-key partitioning across the strip.
+    wikitext_strip_fn = None
+    if policy and policy.get("base_version"):
+        try:
+            from arborist.wikitext import to_base as wikitext_strip_fn
+        except ImportError:
+            wikitext_strip_fn = None
     chunks_for_evidence: list[dict] = []
     total_chars = 0
     for rank, (h, role, per_doc_budget) in enumerate(
@@ -313,6 +326,8 @@ def run_query(
             continue
         r = rows[0]
         span = r.content
+        if wikitext_strip_fn is not None:
+            span = wikitext_strip_fn(span)
         if len(span) > per_doc_budget:
             span = span[:per_doc_budget]
         chunks_for_evidence.append({