feat(03-06): implement literature evidence models, PubMed fetch, and scoring

- Create LiteratureRecord pydantic model with context-specific counts
- Implement PubMed query via Biopython Entrez with rate limiting (3/sec default, 10/sec with API key)
- Define SEARCH_CONTEXTS for cilia, sensory, cytoskeleton, cell_polarity queries
- Implement evidence tier classification: direct_experimental > functional_mention > hts_hit > incidental > none
- Implement quality-weighted scoring with bias mitigation via log2(total_pubmed_count) normalization
- Add biopython>=1.84 dependency to pyproject.toml
- Support checkpoint-restart for long-running PubMed queries (estimated 3-11 hours for 20K genes)

src/usher_pipeline/evidence/expression/__init__.py

@@ -0,0 +1,48 @@
+"""Tissue expression evidence layer for Usher-relevant tissues.
+
+This module retrieves expression data from:
+- HPA (Human Protein Atlas): Tissue-level RNA/protein expression
+- GTEx: Tissue-level RNA expression across diverse samples
+- CellxGene: Single-cell RNA-seq data for specific cell types
+
+Target tissues/cell types:
+- Retina, photoreceptor cells (retinal rod, retinal cone)
+- Inner ear, hair cells (cochlea, vestibular)
+- Cilia-rich tissues (cerebellum, testis, fallopian tube)
+
+Expression enrichment in these tissues is evidence for potential cilia/Usher involvement.
+"""
+
+from usher_pipeline.evidence.expression.fetch import (
+    fetch_hpa_expression,
+    fetch_gtex_expression,
+    fetch_cellxgene_expression,
+)
+from usher_pipeline.evidence.expression.transform import (
+    calculate_tau_specificity,
+    compute_expression_score,
+    process_expression_evidence,
+)
+from usher_pipeline.evidence.expression.load import (
+    load_to_duckdb,
+    query_tissue_enriched,
+)
+from usher_pipeline.evidence.expression.models import (
+    ExpressionRecord,
+    EXPRESSION_TABLE_NAME,
+    TARGET_TISSUES,
+)
+
+__all__ = [
+    "fetch_hpa_expression",
+    "fetch_gtex_expression",
+    "fetch_cellxgene_expression",
+    "calculate_tau_specificity",
+    "compute_expression_score",
+    "process_expression_evidence",
+    "load_to_duckdb",
+    "query_tissue_enriched",
+    "ExpressionRecord",
+    "EXPRESSION_TABLE_NAME",
+    "TARGET_TISSUES",
+]