De-deprecate AtomicCell::get_mut

[jeehoonkang]

PR #332 deprecated after the discussion of issue #315, but it is a
measure not yet necessary to take. Revert the deprecation and wait
for us to reach a consensus.

[ghost]

Since we're not sure whether this is unsound, could we perhaps change the deprecation message to "this method might be unsound and removed in a future release, but we haven't decided yet" and then un-deprecate it if we decide it is sound?

[jeehoonkang]

How about leaving the message and pointing to #315 in the comment, but still de-deprecating the function?

[ghost]

Sure! That sounds fine as well :)

[jeehoonkang]

@stjepang I updated the PR. Thanks for feedback!

[ghost]

Thanks!

bors r+

[bors]

Build succeeded

• continuous-integration/travis-ci/push

[Centril]

Since we're not sure whether this is unsound, [...]

@stjepang If you are not sure this is sound, then I think the right thing to do is to deprecate and eventually remove it until such time that you are sure. In my view, the onus is on you to show why some abstraction using unsafe is sound. Pointing towards the absence of a counter-example is not a good argument. Meanwhile, I think @RalfJung has made a good case for why the soundness is dubious.

[ghost]

@Centril I have reverted this commit and published a new version. Thank you for your feedback!

[Centril]

Thank you!