Support reading and writing indexes
Some references from other tools:
- warcio supports writing indexes from a WARC
- webrecorder's cdxj indexer can write several formats
The classic CDX format is not particularly well-defined nor is it very helpful for indexing beyond "there's a record at this offset". cdxj-indexer can write them and is useful to understand the field semantics. CDXJ is much better-defined and probably more generally useful. It would be good to support reading and writing both.
With index support, we can also support parallel processing of archives provided an index already exists.