GFQL implementation shrink: Cypher lowering core residue investigation

[lmeyerov]

GFQL implementation shrink: Cypher lowering core residue investigation

Reference: #1058

Branch investigated: codex/gfql-lowering-shrink off origin/master @ 2d0be647690d0932306585a857e64bb1cc7cd46c

Target: graphistry/compute/gfql/cypher/lowering.py

Outcome: no PR opened. The safe in-file shrink candidates found in lowering.py did not meet the lane threshold for a meaningful production shrink, so this is the structural receipt path rather than a code PR.

Evidence:

• Enumerated _lower_* helpers and large functions in lowering.py (8,101 LOC on base).
• Cross-checked recent shrink PRs:
  • chore(gfql): shrink Cypher parser handlers #1600 parser-only; no lowering helper made obsolete.
  • chore(gfql): shrink Cypher binder helpers #1602 binder-only; no lowering helper made obsolete.
  • chore(gfql): shrink Cypher projection planning helpers #1608 projection planning helpers already consumed by lowering; no additional safe wrapper deletion found.
  • chore(gfql): shrink Cypher reentry compiletime helpers #1609 reentry compiletime/lowering_support; no lowering.py helper made obsolete.
• No _APPROVED_UNPLANNED_CHAIN_FALLBACK_CODES or equivalent stale approved-unplanned guard remains in current lowering.py.
• Logical-plan defer / optional reentry routing remains active and tested, so it was treated as protected.
• Attempted safe helper extraction around:
  • duplicate projection-name diagnostics,
  • aggregate runtime temp projection wiring,
  • row-column suffix/scope construction,
  • pattern-predicate binding-op lowering,
  • active row-step / UNWIND setup.
• Exploratory patch measured only +399/-428 in lowering.py (net -29) after preserving protected behavior. That is not worth a PR for this lane.
• A candidate dead helper, _rewrite_where_clause_and_resync, is imported by graphistry.compute.gfql.cypher.reentry.compiletime; deleting it broke aggregate reentry tests, so it is not safe deletion residue.

Exploratory validation:

• python3 -m pytest -q graphistry/tests/compute/gfql/cypher/test_lowering.py::test_string_cypher_projection_duplicate_name_error_preserves_context graphistry/tests/compute/gfql/cypher/test_lowering.py::test_lower_match_query_rejects_where_pattern_predicate_introducing_new_aliases passed.
• python3 -m pytest -q graphistry/tests/compute/gfql/cypher/test_lowering.py -k 'aggregate and not cudf' passed (43 passed).
• Local cuDF aggregate slice failed because local host has cuDF installed but no CUDA device; run on dgx-spark only if a future PR touches these runtime paths.

Recommendation:

• Do not open a cosmetic helper-motion PR from this lane.
• Treat lowering.py core aggregate/reentry/defer logic as currently active/protected until adjacent architecture changes land.
• A future meaningful tranche likely needs one of:
  • moving a whole, well-owned lowering subdomain into an existing focused module with net production shrink across files,
  • further retiring direct-Cypher fallback/reentry machinery after its benchmark gates close,
  • or planner-level consolidation that coordinator explicitly authorizes because it touches protected compiler-plan surfaces.