docs(gfql): fix methodology list grouping and scope parity claims

Review follow-ups on PR #1772:
- engines.rst Methodology: the prior-sweep bullet no longer dangles with
  a bare colon over sibling bullets; reworded to state the remaining
  bullets describe that sweep's setup.
- performance.rst + engines.rst: scope the verification claims —
  cross-engine result-row identity applies to the four-engine
  fast-path/index/scaling numbers; the vs-Neo4j and vs-Kuzu pairs were
  validated against expected result rows and cross-database
  value/row-count checks, not cross-engine identity.

Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01Y6dQEcjdazEnzuvuwf73ZL

Author: lmeyerov

docs/source/gfql/engines.rst

@@ -658,9 +658,12 @@ Methodology
 - **0.58.0 release-tag sweep** (the seeded typed-hop snapshot, resident-index and
   flat-scaling numbers, LDBC SNB SF1 / Neo4j 5.26 pairs, and OLAP q8/q9 above): measured
   on the ``0.58.0`` release tag on an NVIDIA DGX Spark (GB10), **warm medians over
-  N=30 runs**, result rows asserted identical across engines before any timing was kept;
-  the Neo4j pairs ran on the same box, warm.
-- Prior bulk sweep (the Orkut / LiveJournal tables and their derived ratios), pre-0.58.0:
+  N=30 runs**. Four-engine numbers were kept only after result rows were asserted
+  identical across engines; the cross-database pairs (Neo4j, Kuzu) were validated against
+  expected result rows and cross-database value/row-count checks. The Neo4j pairs ran on
+  the same box, warm.
+- Prior bulk sweep (the Orkut / LiveJournal tables and their derived ratios), pre-0.58.0 —
+  the remaining bullets in this section describe that sweep's setup:
 - Host: NVIDIA DGX Spark (GB10 Grace-Blackwell, unified memory — the F3 memory-pressure
   boundary is partly a property of this box), RAPIDS container
   ``graphistry/test-rapids-official:26.02-gfql-polars``.

docs/source/gfql/performance.rst

@@ -8,8 +8,9 @@ Engine speedups at a glance
 
 GFQL runs the **same query** on four interchangeable engines — ``pandas`` (default),
 ``polars`` (CPU, columnar), ``cudf`` (NVIDIA GPU), and ``polars-gpu`` (GPU) — and returns
-**identical results** on each (differential parity is a release gate; every number on this
-page was kept only after the result rows were verified identical across engines).
+**identical results** on each (differential parity is a release gate; every four-engine
+number on this page was kept only after the result rows were verified identical across
+engines, and the cross-database pairs were validated against expected result rows).
 Unsupported engine/query combinations are declined before execution during validation,
 compilation, or planning rather than silently falling back. The biggest, easiest win is one
 keyword, **no GPU required**:
@@ -31,8 +32,10 @@ Release-verified numbers (0.58.0)
 ---------------------------------
 
 All numbers in this section were measured on the **0.58.0 release tag** on an NVIDIA DGX
-Spark (GB10), warm medians over N=30 runs, with results asserted identical across engines
-before any timing was kept.
+Spark (GB10), warm medians over N=30 runs. The four-engine numbers (seeded fast paths,
+resident index, scaling) were kept only after the result rows were asserted identical
+across engines; the competitor pairs (vs Neo4j, vs Kuzu) were validated against expected
+result rows and cross-database value/row-count checks.
 
 Seeded typed-hop fast path
 ~~~~~~~~~~~~~~~~~~~~~~~~~~