zooma: batching, retry logic, and ontology filter support

[Copilot]

The zooma tool was processing one row per API call with no fault tolerance, making it brittle for large inputs. This adds batched processing, retry/skip logic, and ontology scoping controls matching the ZOOMA web GUI.

Changes

zooma.py

• Batching: Collects all query values upfront, then sends them in configurable batches (--batch-size, default 10)
• Retry logic:
  • Failed batch → logged to stdout, retried once
  • Second failure → values skipped, logged to stdout
  • Two consecutive batch failures → job fails (server considered unavailable)
• Ontology filter: --ontology-filter (all | efo_obo | obo | custom) + --ontologies for custom comma-separated list; filter is passed as params on annotate requests and as ontologies body field on map requests
• query_zooma_map now accepts a list of values, building the full properties batch array in one call; results matched back to query values by propertyValue field

zooma.xml

• batch_size integer param (default 10, range 1–100)
• ontology_filter conditional with 4 options: All, EFO+OBO, OBO only, Custom; Custom exposes a multi-select (dummy for now — ontology list to be driven by data table in a future release)

macros.xml

• Help text updated to document batching behaviour, retry/skip semantics, and ontology filter options

• Fixes zooma: make tool more robust and flexible #796