Added
- `io`: Read and write support for writing PSMs to Apache Parquet for efficient storage of PSM lists.
- `io.sage`: Support for Sage results in Parquet format (new `SageParquetReader`, renamed `SageReader` to `SageTSVReader`).
Changed
- Upgrade Pydantic dependency to v2. The PSM `spectrum_id` field is now always coerced to a string.
- `io.proteoscape`: Use pyarrow to iteratively read from Parquet instead of first reading an entire dataframe with Pandas.
- `io.sage`: Update compatibility to Sage v0.14
- Remove temporary patch for caching Proforma modification resolvers (now in Pyteomics v4.7.2).