Async version of urdna2015 does not properly batch process hashing in parallel

[dlongley]

The async version of URDNA2015 doesn't take full advantage of multiple processes / thread pooling to process hashes. It was converted some time ago from non-promises to promises and this was probably lost in translation somewhere. When hashing 1st degree quads, for example, each hash call is awaited vs. batching the returned promise and awaiting up to N-many ops at once. This should be fixed to improve performance / make better use of available resources.