fix: Fix NamedEntityExtractor serde

[vblagoje]

Why:

Fixes serialization and deserialization (serde) in the NamedEntityExtractor component. Important for usecases where NamedEntityExtractor needs to be serialized to a pipeline yaml file.

• fixes (De-) Serialization is not properly working for NamedEntityExtractor #7666

What:

• Modified the serialization method to correctly serialize the backend type by calling .name on the _backend.type attribute, ensuring that it is stored as a string rather than an object reference.
• Adjusted the deserialization method to transform the backend string back into the correct NamedEntityExtractorBackend enumeration before attempting to use it in constructing a NamedEntityExtractor instance.
• Included a new test ensuring that the NamedEntityExtractor can be serialized and deserialized correctly, particularly when used within a pipeline configuration.

How can it be used:

• Saving the configuration of a NamedEntityExtractor to a yaml file
• Loading a previously saved NamedEntityExtractor configuration to resume processing or to replicate a specific setup across different environments or applications.

Example of serializing and deserializing a pipeline including a NamedEntityExtractor:

from haystack import Pipeline
from haystack.components.extractors import NamedEntityExtractor, NamedEntityExtractorBackend

# Create an extractor and add it to a pipeline
extractor = NamedEntityExtractor(backend=NamedEntityExtractorBackend.HUGGING_FACE, model="dslim/bert-base-NER")
pipeline = Pipeline()
pipeline.add_component(instance=extractor, name="extractor")

# Serialize the pipeline to a file
with open("path_to_file/test_pipeline.yaml", "w") as f:
    pipeline.dump(f)

# Later or elsewhere, deserialize the pipeline from the file
with open("path_to_file/test_pipeline.yaml", "r") as f:
    loaded_pipeline = Pipeline.load(f)

How did you test it:

• Updated the unit test for the NamedEntityExtractor to include checks for both serialization and deserialization functionality, confirming that the process retains the correct component configurations.
• Added a new test that incorporates the NamedEntityExtractor within a pipeline configuration, ensuring that the entire pipeline can be serialized and deserialized correctly. This test verifies that the pipeline's dictionary representation remains consistent before and after the serde.

Notes for the reviewer:

• Pay special attention to the changes in the serialization and deserialization methods, ensuring that the implementation aligns with standard practices for handling enumeration types in Python.

[coveralls]

Pull Request Test Coverage Report for Build 9076600918

Details

• 0 of 0 changed or added relevant lines in 0 files are covered.
• 20 unchanged lines in 1 file lost coverage.
• Overall coverage increased (+0.02%) to 90.423%

---

Files with Coverage Reduction	New Missed Lines	%
components/extractors/named_entity_extractor.py	20	68.67%

Totals
Change from base Build 9074873557:	0.02%
Covered Lines:	6534
Relevant Lines:	7226

---

💛 - Coveralls

[davidsbatista]

I'm a bit confused; where is this init_params used? it it's never used inside this function

[davidsbatista]

OK, nevermind, I understood now what's going on :)

[davidsbatista]

LGTM

[vblagoje]

@davidsbatista there was a conflict with main branch commit that I needed to resolve. I also took the opportunity to remove unit test markers as we don't use them any more. Please double check again everything 🙏