Speed up temporary shape deduction by using a batched access range mapper call (closes #13). See merge request !61