v2.9.8

Fix #335 (part 1): wire GENCODE_BIOTYPE_ALIASES from gtfparse 2.7.0 into pyensembl's read_gtf call. GENCODE GTFs (which use gene_type / transcript_type) now get those columns renamed to the Ensembl canonical gene_biotype / transcript_biotype at parse time, so Transcript.is_protein_coding and biotype-filtered queries work without a manual rename pass. Bumps gtfparse dep floor to >=2.7.0. Combined with v2.9.6 (versioned protein-ID FASTA matching), this closes the original #335 GENCODE-genome repro end-to-end.