Use the snapshot cycle first block level to get the block.

[novalis]

With short cycles, the current level might be in a different cycle, causing the RPC to fail.

• Analysis: I am trying to make tezos-reward-distributor work with private Tezos chains. During testing, I set my blocks_per_cycle to 8 (instead of the normal 8192). This makes testing much quicker, as one only needs to wait a few minutes to get enough blocks baked to do a distribution. But I noticed that requests were failing. One such failing request was to http://my-private-tezos-node/chains/main/blocks/14396/context/selected_snapshot?cycle=1795 I noticed that block 14396 should not be in cycle 1795 (with a cycle length of 8). This led me to this funny bit of code.

• Solution: For a parameter named snapshot_cycle_first_block_level, use a variable named snapshot_cycle_first_block_level not current_cycle

• Performed tests: After this change, tezos-reward-distributor gets further in distributing rewards (it still doesn't work, but that might be a config issue on my side).

Work effort: 3 hours (mostly spent learning what any of this code was doing, and figuring out how to get reasonable error messages -- see #597).

[nicolasochem]

Thanks! I tried this against mainnet RPC for one mainnet baker as a dry run and your branch and ithaca2 gave me the same calculation csv file, so I'm merging into ithaca2.

@jdsika we should really look into merging ithaca2 into master.