[+] add python based iceberg sink

[0xgouda]

• The server assumes a PostgreSQL catalog is used and creates pgwatch namespace and pgwatch.metrics table within it if they don't exist.
• The table is partitioned by MetricName and DBName (in order).
• Metrics are written in the local file system as Apache Arrow records with the following schema:

  Schema(
      NestedField(field_id=1, name="DBName", field_type=StringType(), required=True),
      NestedField(field_id=2, name="MetricName", field_type=StringType(), required=True),
      NestedField(field_id=3, name="Data", field_type=StringType(), required=True),
  )

• Catalog configurations should be provided in .pyiceberg.yaml file under pgcatalog.

---

Note

This PR focused on providing a working PoC for a Python-based Iceberg receiver, so there are still a couple of Todos that may be added in the future to make this receiver more production-ready:
1. Use object storage instead of the local file system.
2. Support TLS over the gRPC connection.
3. Add authentication interceptor.
4. Cache measurements to minimize the number of Parquet files written.

[destrex271]

awesome work!