Requests to delete data should be handled asynchronously

Delete requests are currently handled synchronously within the http request thread. 9 out of 10 times the delete operation will complete before the http timeout hits, but in cases of large data set it can become an issue that the delete operation exceeds the http timeout window and the transaction is rolled back resulting in undeletable data.

Delete requests for sites, jobs and builds should be handled asynchronously to the http request to allow the database transaction to complete even if it takes longer than the http request timeout window allows.