Fix/update cropus name embeddings model lang

[lpi-tn]

This pull request updates the materialized view for corpus embedding models to include additional metadata and ensures the corresponding SQLAlchemy model and versioning are in sync. The core change is a migration that drops and recreates the corpus_name_embedding_model_lang materialized view with more fields and improved logic, and updates the data model accordingly.

Database migration and schema changes:

• Added a new Alembic migration (b049924f7067_modify_corpus_name_embedding_model_lang_) that drops and recreates the corpus_related.corpus_name_embedding_model_lang materialized view. The new view now includes corpus_id, embedding_model_id, used_since, and category_id, and ensures only the latest embedding model per corpus and language is kept using a window function.
• Updated the CorpusNameEmbeddingModelLang SQLAlchemy model in corpus_related.py to add new fields: corpus_id, embedding_model_id, used_since, and category_id, matching the new view schema.

Versioning:

• Bumped the package version in pyproject.toml from 1.4.0 to 1.4.2 to reflect the schema and model changes.

[Copilot]

Pull request overview

This PR updates the corpus_related.corpus_name_embedding_model_lang materialized view to expose additional metadata (corpus/model IDs, used_since, category_id) and keep only the latest embedding model per (corpus, language), then aligns the SQLAlchemy read-only model and bumps the package version.

Changes:

• Recreates the corpus_name_embedding_model_lang materialized view with extra columns and a ROW_NUMBER()-based “latest per corpus/lang” selection.
• Extends the CorpusNameEmbeddingModelLang SQLAlchemy model to match the new view schema.
• Bumps project version from 1.4.0 to 1.4.2.

Reviewed changes

Copilot reviewed 3 out of 3 changed files in this pull request and generated 1 comment.

File	Description
welearn_database/data/models/corpus_related.py	Adds new fields to the ORM model representing the updated materialized view.
welearn_database/alembic/versions/b049924f7067_modify_corpus_name_embedding_model_lang_.py	Drops/recreates the materialized view with updated projection and “latest per corpus/lang” logic.
pyproject.toml	Version bump to reflect the schema/model change.

---

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.

[Copilot]

source_name is declared as the sole primary key, but the materialized view returns one row per (corpus_id, lang) (latest used_since), which can produce multiple rows for the same source_name across different languages. With source_name alone as the ORM PK, SQLAlchemy’s identity map can collapse/overwrite rows and return incomplete/incorrect results. Consider using a composite primary key that matches the view’s uniqueness (e.g., include lang and/or corpus_id).

Suggested change

	source_name: Mapped[str] = mapped_column(primary_key=True)
	corpus_id: Mapped[UUID]
	embedding_model_id: Mapped[UUID]
	title: Mapped[str]
	lang: Mapped[str]
	source_name: Mapped[str] = mapped_column()
	corpus_id: Mapped[UUID] = mapped_column(primary_key=True)
	embedding_model_id: Mapped[UUID]
	title: Mapped[str]
	lang: Mapped[str] = mapped_column(primary_key=True)