feat(bulk-load): replica handle errors during bulk load

[hycdong]

As previous pull requests shows, bulk load process is like:

1. meta sends bulk load request to primary replica to let primary replica executing bulk load (feat(bulk-load): meta server send bulk load request #457)
2. primary replica broadcasts group_bulk_load_request to secondaries to let secondary replicas executing bulk load (feat(bulk-load): group bulk load request in replica group #460)
3. secondary replica reports its bulk_load_state through group_bulk_load_response (feat(bulk-load): bulk load download part4 - replica report download status and progress #479)
4. primary replica reports group bulk_load_state through bulk_load_response (feat(bulk-load): bulk load download part4 - replica report download status and progress #479)

This pull request is about how replica handle errors during bulk load process, including two parts:

1. primary meet errors while receiving group_bulk_load_response
   • primary stores secondary bulk load state in primary context, when meet errors, primary should reset this secondary's state (in function on_group_bulk_load_reply and reset_node_bulk_load_states)
2. replica partition status changes during bulk load
   • if primary changes, primary should clean up its primary bulk load states in primary context (feat(bulk-load): bulk load download part1 - replica start download #465 feat(bulk-load): bulk load download part4 - replica report download status and progress #479 feat(bulk-load): bulk load ingestion part5 - replica handle bulk load request during ingestion #496)
   • if other status switches, bulk load states should also be cleared (in function update_local_configuration and clear_bulk_load_states_if_needed)