Implements hashing as a blanket trait for instances of CanonicalSerialize

[huitseeker]

• this saves an alllocation w.r.t the issue-suggested approach by implementing io::Write on the input instance of digest::Digest,
• note that most instances of digest::Digest already have an io::Write instance, but CanonicalSerialize consuming its io::Write argument prevents its usage,
• this hence implements io::Write on a cheap newtype wrapper

closes: #86

---

• Targeted PR against correct branch (master)
• Linked to Github issue with discussion and accepted design OR have an explanation in the PR that describes this work.
• Wrote unit tests
• Updated relevant documentation in the code
• Added a relevant changelog entry to the Pending section in CHANGELOG.md
• Re-reviewed Files changed in the Github PR explorer

[Pratyush]

but CanonicalSerialize consuming its io::Write argument prevents its usage,

Is it not possible to work around this by doing &mut writer (&mut W is Write whenever W is)

[huitseeker]

@Pratyush Sure, though this signature change would change the API of CanonicalSerialize in a backwards-incompatible way.

Moreover, I'm not sure we want to depend on the std::io::Write implementations in RustCrypto/hashes:

• the Digest trait does not guarantee its existence on its own,
• it's implemented from unrelated (MAC) trait impls through a macro which future in the code base isn't assured.
• the no-std context adds an additional wrinkle.

They're here .. just not very reliably so.

[Pratyush]

Ok that makes sense, let's proceed as is. This LGTM modulo the change to CanonicalSerializeHashExt

[Pratyush]

Oh @huitseeker if you could add a CHANGELOG entry that would be great (also for #263)

[ValarDragon]

Thanks for adding this!