Revert Bytes implementation changes introduced by #95

[alexandruag]

#95 fixed the problem of torn reads/writes caused by the implementation of read/write_obj essentially leveraging the environment specific memcpy, and the implicit assumption that read/write_obj perform atomic accesses up to a certain size at aligned addresses. Meanwhile, we added the load/store operations to Bytes which provide explicit atomic access semantics. We should document that read/write_obj must not be used when atomicity is required, and revert the related changes from #95.

Here are some next steps we should consider:

• Explicitly document the semantics of read/write_obj with respect to atomicity.
• Reach out to folks using vm-memory and/or discuss here about the implications of the transition.
• Consumers of the crate have to replace read/write_objs that are supposed to be atomic with load/stores. Let's do a 0.3.x (or later) release right before removing anything, to help with the transition.
• Apply the changes and include them in a 0.4.x (or later).