new operation_id generation scheme for hive.operations

operation id is now created with encoding block num, operation type and operation number in block into 64 bit value. This allows to apply several opts:

• try to lookup for block-specific operations by specifying operation_id between <block_num, 0, 0> and <block_num, 0xFF, 0xFFFFFF>
• same could be done by composing op_type_id into constant above
• then we can try remove block_num, op_type_id fields from table storage and define them only at access views like hive.operations_view, hive.account_operations_view and all app-context specific views. The drawback is lacking of virtual column support in Postgres (which exists but really requires a storage what is a nonsense)

This also solves a problem related to nondeterministic values of operation-ids which now finally can have different values for each HAF instance depending on forks (because old solution is always incrementing a sequence providing values for operation-id)