Add bidirectional OpenMetadata semantic model converter (converters/openmetadata)

[daniyarndh]

OpenMetadata has emerged as a widely adopted, enterprise-grade unified metadata catalog. It supports rich business glossaries, metrics definitions, and lineage tracking. However, there is currently no standardized way to serialize these cataloged assets into a vendor-neutral, developer-friendly, and version-controlled format that BI engines and AI agents can consume.

By building a bidirectional converter, we bridge OpenMetadata's central governance platform with Apache Ossie's declarative semantic standard, turning OpenMetadata into a first-class participant in the Ossie ecosystem.

Proposed Design & Mapping

The converter will be written in Python, matching the structure of existing converters like converters/dbt and converters/databricks.
Attributes specific to OpenMetadata that are not representable in the core Ossie specification (e.g., column-level tags, owners, tiers, and domain classifications) will be preserved in a custom_extensions block using the OPENMETADATA vendor namespace:

custom_extensions:
  - vendor_name: OPENMETADATA
    data:
      owner: "finance_team"
      tier: "Tier1"
      tags: ["PII.Sensitive", "Finance.Reporting"]

The converter package may also ship a CLI, e.g.:

# Export OpenMetadata to Ossie YAML
osi-openmetadata import --api-url http://localhost:8585 --jwt-token <token> --glossary "Finance" -o finance_semantic_model.yaml

# Ingest Ossie YAML back into OpenMetadata
osi-openmetadata export -i finance_semantic_model.yaml --api-url http://localhost:8585 --jwt-token <token>