MIRI workarounds possibly not needed?

[nico-abram]

Since rust-lang/miri#1952 MIRI understands the Unpin workaround, so maybe the MIRI cfg is no longer needed.

That said, to quote RalfJ from that PR:

But can an &Aliasable or &mut Aliasable coexist with an &mut T (or to a field of T)?

&mut T (with this patch: for T: Unpin) cannot coexist with anything (assuming both pointers are being used). Weakening that guarantee would make it entirely useless for optimizations.

But &mut T can be reborrowed from &mut Aliasable, as long as for the time that the &mut T is active, no other reference/pointer to that memory is used.

Which seems to contradict the readme:

so that one &mut Aliasable and either one &mut T or any number of &Ts can coexist

[SabrinaJewson]

Thanks for noticing this, I've removed the workaround now - it seems that change was made after I intially wrote this crate.