Add option to use existing CDXJ indices rather than indexing from WARCs

[tw4l]

Related to webrecorder/browsertrix-crawler#484

In Browsertrix Crawler, we are already generating CDXJ indices per-WARC, so it would be faster to use these existing indices rather than indexing from the WARCs again. We are proposing adding a --cdxj CLI argument that can pass a directory of existing CDXJ files, similar to how --pages already works.

I have a PR in progress, just needs a bit more testing, will submit shortly. Thanks!