make faster reaching head block by an application after its break

When a HAF application is working on the blockchain head its context's irreversible block is always behind a current block num. When for some reason an application breaks its stuck for a while, then after re-run it will process block one by one instead of move to massively processing all irreversible blocks that were added during pause. This happens because of 'hive.app_next_event' processes all the NEW_IRREVERSIBLE events and its current block is always before its irreversible block, so it is impossible to return a range of blocks to process.

The solution is to set a context's irreversible block to the biggest possible irreversible block -> biggest known irreversible block num less than a fork event block_num before a context (if any fork is ahead)