fix: Preserve GLiNER config fields through Pydantic validation

[yuriihavrylko]

Change Description

Add GLiNERRecognizerConfig to the YAML recognizer config model and register it in CONFIG_MODEL_MAP, so that model_name and other GLiNER-specific fields (flat_ner, threshold, multi_label, map_location, load_onnx_model, onnx_model_file, entity_mapping) are preserved through Pydantic validation when loading recognizer configuration from YAML.

Root cause: When parsing a GLiNERRecognizer entry from YAML, parse_recognizers looked up "GLiNERRecognizer" in CONFIG_MODEL_MAP and found nothing, falling back to PredefinedRecognizerConfig. Since that class has no model_name field and Pydantic's default is to ignore extra fields, model_name (and all other GLiNER params) were silently dropped. The recognizer would then always load the hardcoded default model (urchade/gliner_multi_pii-v1) regardless of what was specified in config. HuggingFaceNerRecognizer already had this handled correctly via HuggingFaceRecognizerConfig.

Issue reference

Fixes #XX

Checklist

• I have reviewed the contribution guidelines
• I have signed the CLA (if required)
• My code includes unit tests
• All unit tests and lint checks pass locally
• My PR contains documentation updates / additions if required

[yuriihavrylko]

@microsoft-github-policy-service agree