Support reading datasets from a custom block

[kalabukdima]

Right now, the scheduler always tries to read all blocks starting from 0. If blocks in a bucket start from some other height, this bucket is ignored with the following error message:
Error getting data chunks: Blocks 0 to 177455579 missing from neon-devnet-traceless-1

• Don't expect blocks to start from 0 in continuity check
• Allow setting custom start height to explicitly ignore some existing chunks