continue as new for retry should have reason

[yiminc-zz]

When workflow is scheduled with retry policy, and when the workflow failed server would do continue as new instead of fail the workflow and a new workflow execution will be started. The last event of the failed workflow execution will be ContinueAsNew event. When user look at the event they have no clue what caused the failure. We should add a FailedReason/FailedDetails to the ContinueAsNew event so when people look at the event it knows why it is doing continue as new.