ImportDocument: prevent retry loops on duplicate-result entity lookups during PDF ingestion

[Copilot]

ImportDocument can get stuck reprocessing some PDFs when entity resolution fails with backend error Id loading expect only one response (notably on Vulnerability / Attack-Pattern lookup paths). This change hardens lookup behavior so the connector can continue processing instead of failing the message and looping.

• Lookup hardening for parsed entities

  • Added DUPLICATE_ID_ERROR_MESSAGE and _read_entity_with_fallback(...) in reportimporter/core.py.
  • Vulnerability (Vulnerability.name) and MITRE technique (Attack-Pattern.x_mitre_id) resolution now use this helper.

• Controlled fallback behavior

  • Keep normal path as read(filters=...).
  • Only for the known duplicate-result backend error, fallback to list(getAll=True, filters=...).
  • Re-raise unexpected exceptions (with contextual warning) to avoid masking unrelated failures.

• Deterministic and safe fallback selection

  • Handle list() returning None.
  • Drop invalid list results without id (with warning).
  • Deterministically select first entity after sorting by id.
  • Emit explicit warnings for multi-match and empty-usable-result cases.

# core idea introduced in ImportDocument entity resolution
try:
    entity = api_entity.read(filters=filters)
except Exception as e:
    if DUPLICATE_ID_ERROR_MESSAGE not in str(e):
        raise
    entities = api_entity.list(getAll=True, filters=filters) or []
    entities = [x for x in entities if x.get("id")]
    entity = sorted(entities, key=lambda x: x["id"])[0] if entities else None