To make the library work as `no_std` we add an `std` feature which
is on by default. When it is off, we compile as `no_std` and make
parts of the API that require `std::io` conditional on the `std`
feature.

As consequence Buf::collect is removed as well, which is replaced with `Buf::into_bytes`. The advantage of `Buf::into_bytes` is that it can be optimized in cases where converting a `T: Buf` into a `Bytes` instance is efficient.

- Clones when the kind is INLINE or STATIC are sped up by over double.
- Clones when the kind is ARC are spec up by about 1/3.

The previous implementation didn't factor in a single `Bytes` handle
being stored in an `Arc`. This new implementation correctly impelments
both `Bytes` and `BytesMut` such that both are `Sync`.

The rewrite also increases the number of bytes that can be stored
inline.