feat: add DuckLake datasource

[anoop-narang]

For #120

Summary

• Add DuckLake as a new datasource using the datafusion-ducklake crate (v0.0.6)
• DuckLake stores metadata in a Postgres catalog database and data as Parquet files on S3
• Implements table discovery via CatalogProvider/SchemaProvider traversal and data fetch via DataFusion SessionContext
• Credentials passed as a single JSON blob (catalog_db_password + optional S3 keys)

Usage

curl --location 'http://localhost:3000/connections' \
--header 'Content-Type: application/json' \
--data-raw '{
  "name": "my_ducklake",
  "source_type": "ducklake",
  "config": {
    "catalog_url": "postgresql://<user>@<host>:<port>/<database>?sslmode=require",
    "s3_endpoint": "https://s3.<region>.amazonaws.com",
    "s3_region": "<region>",
    "credentials_json": "{\"catalog_db_password\": \"<password>\", \"s3_access_key_id\": \"<access_key>\", \"s3_secret_access_key\": \"<secret_key>\"}"
  }
}'