Add GRCh38/CHM13 CpG-island and GENCODE gene reference fixtures for DMR

[YannVRB]

What

Reference annotation data for IntGenomicsLab/lrsomatic's new haplotype DMR feature (IntGenomicsLab/lrsomatic#204): CpG island coordinates (to restrict DMR calls to) and gene models (to annotate DMR calls with their nearest gene), for both GRCh38 and CHM13.

Files added

• references/dmr/{GRCh38,CHM13}.cpg_islands.bed.gz — genome-wide CpG islands, for production use as lrsomatic's igenomes.config defaults.
• references/dmr/{GRCh38,CHM13}.gencode_gene.bed.gz — genome-wide gene models, same purpose.
• references/dmr/{GRCh38,CHM13}_chr19.{cpg_islands,gencode_gene}.bed — chr19-only, uncompressed subsets for CI (matches lrsomatic's existing chr19 test profile).

Sources

• GRCh38 CpG islands: UCSC cpgIslandExt.txt.gz (bin column stripped).
• GRCh38 gene models: GENCODE release 48 annotation GTF, gene-feature rows only.
• CHM13 CpG islands: UCSC's T2T-CHM13v2.0 bigBed track (GCA_009914755.4_T2T-CHM13v2.0.cpgIslandExt.bb), converted via bigBedToBed, then remapped from GenBank accession contig names (e.g. CP068259.2) to chr-style names via UCSC's own chromAlias table for this assembly.
• CHM13 gene models: UCSC's catLiftOffGenesV1.gtf.gz (CAT+Liftoff T2T-CHM13v2.0 annotation), which has no gene-level features -- gene extents computed by grouping transcript rows by gene_id (min start, max end).

All chr19 subsets are position-sorted (bedtools closest, used downstream, requires sorted input).