Sharders finalized different block in dev network

[peterlimg]

In dev network, dev1 finalized different block, check how it happened.

[peterlimg]

See details in slack thread

Peter: The main issue is that sharders were in different LFBs when they get restarted, so if sharder 1, 2, 3 are with LFB 100, 101, 102, miners will fetch LFB from sharders when they get restarted, so if sharder 3 that has highest LFB round number get up the last, then miners may get LFB as round 100 from sharder1 (or 101 from sharder2). The problem is they will start from the LFB and regenerate blocks for the following rounds. And in our case, the miners generated new block for 102 and finalized it, both sharder 1 and 2 accept, but sharder 3 will see different block for the 102 as it has LFB of round 102 stored in local state before the restart.