Clinlp

Added
- a custom component for entity recognition, with options for proximity, fuzzy and pseudo matching

Added
- definition for qualifiers (negation, plausibility, temporality, experiencer)

Changed

- updated rules for context algorithm to be consistent with definitions
- added some rules for context algorithm
- refactored `Qualifier` class from enum to a separate class, that accomodates other fields (like prob)
- use `entity._.qualifiers` to obtain `Qualifier` classes, `entity._.qualifier_str` for strings, and `entity._.qualifier_dict` for dicts

Fixed

- ambiguity of `dd` for context rules (can mean differential diagnosis, and daily dosage)
- importing `clinlp` caused a bug when extras were missing

Removed
- support for python 3.9

Added

- remove a default spacy abbreviation (`ts.`)
- option for max scope on qualifier rules, limiting the number of tokens it applies to
- a transformer based pipeline for negation detection (`clinlp_negation_transformer`)
- a base class `QualifierDetector` for qualifier detection

Fixed

- issue where entity and context trigger were overlapping (e.g. `geen eetlust`)
- some tests that were not auto-discovered by pytest due to naming

Changed
- refactored context algorithm to allow adding new qualifier detectors
- the `clinlp_autocomponent` wrapper as a utility function, which makes creating components with inheritance and arbitrary config a bit easier
- made default configs a bit simpeler and DRY
- move qualifier adding for context algorithm to base class

Added

- version info to model meta (warns if installed `clinlp` version does not match model version)
- a component for normalizing

Fixed

- bug with resource loading