add option to consolidate a dataset

[banteg]

a freshly collected contracts dataset out of cryo is 15.53 gb. if you consolidate 17,920 files into 17 files, it would become 7.88 gb, providing 2x savings on storage and 3x improvement in query performance.

i propose to add an option for cryo to incrementally consolidate the collected datasets, merging parts as a soon as they form a larger chunk that won't require any further rewriting.

an example of how it could work with --align option:

• 0-17,000,000 block range is consolidated into files of 1,000,000 blocks each
• 17,000,000-17,900,000 range is consolidated into files of 100,000 blocks each
• 17,900,000-17,920,000 range is consolidated into files of 10,000 blocks each
• 17,920,000-17,924,000 range is kept as collected with 1,000 blocks in each file
• if we run again after block 17,930,000, blocks 17,920,000-17,930,000 would be consolidated into a bigger file
• same would happen block 18,000,000 with chunks for block 17,900,000-18,000,000

adopting this approach would allow a set-and-forget cron job with cryo <dataset> --align --consolidate for a researcher to always come back to a fresh and performant dataset.

[banteg]

implemented the logic i described here
https://github.com/banteg/cryogen