Support unaligned load/store on more ARM targets

[Anton Korobeynikov]

> The patch looks good to me, except that AllowsUnalignedMem is not guaranteed
> to be false by that point on subsequent calls. It was a fault on the
> previous code as well, I guess. I wouldn't save the "else false" in the
> default case.
>
> Anton, any comments?
Generally, looks good to me. It seems that indeed the target feature
is the best solution here.

--
With best regards, Anton Korobeynikov
Faculty of Mathematics and Mechanics, Saint Petersburg State University