Fix biotype-key annotation on master

[byee4]

Summary

• Bring the tested biotype-key fallback from Handle alternate GTF biotype keys #9 onto master. PR Handle alternate GTF biotype keys #9 was merged into the branch for Test Ensembl transcripts without gene features #8, so its commit was absent from master.
• Recognize both gene_type/transcript_type and gene_biotype/transcript_biotype when a GTF format setting and its actual keys differ.
• Include coding and noncoding regression tests, plus README and changelog updates.

Verification

• uv run --no-project --with pytest --with gffutils --with future --with tqdm --with pybedtools --with pandas --with numpy python -m pytest -q unit_tests — 12 passed; CI passed.
• I tested all 168 BED rows attached to Only noncoding annotations returned despite protein-coding genes in reference DB #6 against a gffutils DB built from all GENCODE v49 features for the overlapping genes. With --format gencode, the original master and this branch give identical results: 112 3utr, 26 CDS, 10 5utr, and 20 intronic calls, all assigned to protein-coding genes.
• With --format ensembl, the original master reproduces the issue's noncoding-only labels. This branch restores coding classification for CDS and introns under that mismatched format, but GENCODE's generic UTR features still require --format gencode for 5′/3′ classification.

Related to #6. The issue remains open while the reporter's exact invocation is checked.