BREAKING CHANGE: v0.0.8 -> sort/merge --rename impl + merge preserves unique metadata lines

[alejandrogzi]

[0.0.8] - 2025-06-13

Added

• --rename/-r flag for merge — chaintools merge --rename reassigns chain
  IDs sequentially in sorted output order, so the first emitted chain receives
  id 1, the next 2, and so on. The flag implies sorting: it defaults to
  --sort-by score when no key is given, and respects an explicit --sort-by <KEY>
  when one is provided. Renaming is applied at final emission for both the in-memory
  and external (spill-to-disk) merge paths, with no extra pass over the data. Backed
  by new write_chain_dense_with_id / write_chain_header_with_id writer helpers.

• --rename/-r flag for sort — chaintools sort --rename reassigns chain
  IDs sequentially in sorted output order, following the selected --sort-by key
  (default score). When combined with --out-index, the recorded offsets are
  computed against the renamed output so they remain consistent with the written
  bytes.

Changed

• merge deduplicates metadata — when merging with --sort-by/--rename,
  identical metadata (#) lines that appear in more than one input (for example,
  files produced by a previous split) are now emitted once, in first-seen order.
  Deduplication is performed at the byte level and does not affect sort, which
  continues to preserve a single input's metadata verbatim.