Embedding fixes

[NiklasAbraham]

Added small fixed in embedding and mutation and standard numbering.

[Copilot]

Pull Request Overview

This PR introduces a normalize flag throughout embedding calculation routines, standardizes Neo4j queries to use elementId instead of id, and updates several dependencies and logging patterns.

• Adds normalize: bool = True parameter to all embedding methods and propagates it through the processor and model implementations
• Replaces id(r) with elementId(r) in standard numbering and mutation detection queries
• Improves logging in sequence alignment and ontology loading, and updates project dependencies

Reviewed Changes

Copilot reviewed 11 out of 11 changed files in this pull request and generated 3 comments.

Show a summary per file

File	Description
src/pyeed/embeddings/processor.py	Added normalize parameter to public/legacy embedding APIs
src/pyeed/embeddings/models/prott5.py	Propagated normalize and pooling changes for ProtT5
src/pyeed/embeddings/models/esmc.py	Added normalize flag to ESMC batch/single embedding
src/pyeed/embeddings/models/esm3.py	Added normalize flag to ESM3 batch/single embedding
src/pyeed/embeddings/models/esm2.py	Added normalize flag to ESM2 batch/single embedding
src/pyeed/embeddings/base.py	Updated abstract methods to include normalize parameter
src/pyeed/analysis/standard_numbering.py	Switched to elementId and refined logging
src/pyeed/analysis/sequence_alignment.py	Replaced print with logger, fixed query patterns
src/pyeed/analysis/ontology_loading.py	Improved OWL restriction handling and relationship logic
src/pyeed/analysis/mutation_detection.py	Updated region_ids_neo4j type and elementId queries
pyproject.toml	Removed old numpy constraint; replaced umap with umap-learn

Comments suppressed due to low confidence (1)

src/pyeed/embeddings/models/prott5.py:145

• The variable attention_mask is not defined in this scope. You need to obtain it from the model outputs (e.g., outputs.attention_mask) or pass it into the method.

        seq_len = attention_mask.cpu().numpy().sum()