I looked through the commits and this literally just looks like interface fluff to get it to recognise the M1 as an OpenCL capable device.
Given relatively little is known about the architecture of the M1 GPU it's not surprising that it's only getting 2MH/s. The NVidia and AMD GPU miners have had a considerable amount of time and skill investment over the years to get the rates we see now and that's with considerably more knowhow into how the GPUs work. I'd imagine the 60GB/s or so of bandwidth is a limiting factor, and whilst I'm sure apples memory controller is fantastic, it's shared by system processes* and dwarfed by even el cheapo 128-bit GDDR5 cards, but I reckon it's capable of more though we'll likely not see it, because really why bother.