feat: mztab support

[cmdoret]

Context:

mzTab is a commonly used file format to represent results from mass spectrometry experiments. As we plan to support proteomics and metabolomics data, we need to enable metadata extraction and data access for this file format.

Changes

• pin modos-schema to metabolomics branch
• add parsing utilities to extract metadata into modos schema from mztab contents
• implement logic to populate zarr metadata from mzTab
• add a modos enrich command to trigger metadata extraction from file contents
  • at first, I thought of putting it as a hook whenever something is added, but given that it can be very slow, one may not want to run it right away.
• all relative imports made absolute (import .helpers -> import modos.helpers)
• Add boilerplate for extraction of arrays from file contents into the zarr hierarchy
  • basic implementation that nests the arrays directly in the parent group

Limitations

• We rely on pyteomics for mzTab parsing, which comes with a large dependency tree (inc. sklearn)
• differences in mztab between proteomics (mztab 1.0) and metabolomics (mztab 2.X-M) need to be clarified
  • this PR focuses on mztab 2.X-M
• The file is included as-is and only the metadata is can be accessed through s3 using zarr. If relevant, the next step would be to extract data tables into zarr arrays and provide the same flexibility.

Notes
Exploratory tests with array extraction from mztab files revealed little benefit of exposing arrays directly in zarr:

• zarr supports multi-dimensional arrays, not column-based lists -> better for fixed-size and (mostly) homogenous datatypes
  • it appears that it is currently not possible to retain both column names (structured datatypes, see zarr tuto) and use variable-size datatypes (arbitrary length strings, see object arrays)
• mztab arrays are composed of mostly strings and variable datatypes
• even integer id columns are strings, as they can contain mulitple identifiers separated by |
• tables are relatively small, making zarr features less interesting