DRY: consolidate duplicated DB/graph helpers (_sanitize_value, _formatResultAsMarkdown, build-loop) into ai.common.utils

[dsapandora]

Problem Statement

@asclearuc's review on #1584 flagged several near-duplicate helpers across the DB/graph nodes that have already drifted:

• _sanitize_value / _sanitize_row exist in db_instance_base and graph_instance_base with divergent behaviour (Decimal → 1.5 vs '1.50' before feat(nodes, ai): add graph base class; make FalkorDB a real graph node #1584 aligned the graph copy).
• _formatResultAsMarkdown has ~5 near-verbatim copies (db_instance_base, db_arango, db_neo4j, graph_instance_base) with silent per-copy differences (e.g. a missing (headers, rows) tuple branch).
• The _buildQuery / _buildSQLQuery retry loop is duplicated — and the graph fix is not in the SQL one (see the companion bug issue).
• parse_is_valid was removed in feat(nodes, ai): add graph base class; make FalkorDB a real graph node #1584, but the SQL base still carries a third .lower() == 'true' variant of the same idea.

Proposed Solution

Extract the genuinely engine-agnostic helpers — _sanitize_value, _formatResultAsMarkdown, and the build-loop control flow — into ai.common.utils, and consume them from db_instance_base, db_arango, db_neo4j, and graph_instance_base. Keep the engine-specific driver-level serialisers local (they legitimately differ per engine). Do this as a dedicated refactor pass, not smuggled into a feature PR.

Alternatives Considered

Leave the copies as-is — rejected: they have already diverged silently and will keep doing so as each node evolves.

Affected Modules

• nodes (pipeline)
• ai

---

DRY follow-up to #1584. Lower priority than the correctness port (companion bug issue).

[github-actions]

🤖 Internal: Discord sync marker

Auto-managed by the Discord notification workflow. Stores the linked Discord message ID and forum thread ID. Do not edit or delete.