Feature Request: Metadata Compatibility Challenge Between DuckLake and Delta/Iceberg Formats

[djouallah]

Delta Lake and Apache Iceberg both assume that the atomic unit of a snapshot is a complete Parquet file. While DuckLake offers greater flexibility in this regard, this design difference creates a fundamental compatibility issue: there is no straightforward mechanism to generate Delta Lake or Iceberg metadata that references partial Parquet files.
This architectural mismatch means that DuckLake's flexible snapshot approach cannot be easily mapped to the metadata structures expected by Delta and Iceberg formats, limiting interoperability between these systems.

for example data inlining is easy to fix, just flash the data, parquet deletion, just rewrite the file, but I am stuck on partial file snapshot