feat: add digest-traits feature to implement digest:Update for all hashers

[hanabi1224]

This PR

• add digest-traits feature to implement digest::Update for all hashers
• bump dep versions

digest::Update is required by digest_io::IoWrapper which is useful for streaming input to a hasher.

e.g.

           // `digest-io` compatibility
           #[cfg(feature = "digest-traits")]
           {
                let mut hasher = digest_io::IoWrapper(<$alg>::default());
                let mut reader = std::io::Cursor::new($data);
                std::io::copy(&mut reader, &mut hasher).unwrap();
                assert_eq!(
                    $code.wrap(hasher.0.finalize()).unwrap().to_bytes(),
                    expected,
                    "{:?} encodes correctly (from hasher)", stringify!($alg)
                );
           }

[vmx]

Thanks for the PR, the changes look good. I'd like to have the upgrade to digest 0.11 in a separate commit, so I just went ahead and did it myself at #406 (a review would be welcome, we are stretched thin on reviewers).

In regards to your change. I don't think we need to feature gate this. Making digest a non-optional shouldn't causes any problems.

Do I understand correctly that the digest-io test is testing that digest::Update is actually implemented? If yes, please add a comment that makes that clearer (at a first look was wondering why there are no tests for it).

[hanabi1224]

@vmx Thanks for your quick feedback! If we don't feature-gate this, should I go a step further to implement digest::Update for T: multihash_derive::Hasher in multihash_derive crate? That would make the implementation automatic for new hashers, but it introduces 1 more required dependency to the multihash_derive crate. What do you think?

[vmx]

@hanabi1224 Good point. Implementing it directly in multhash_derive makes sense. New dependencies indeed always need a second thought, but as digest only contains traits it's not introducing much bloat.

[hanabi1224]

@vmx PR updated.
(BTW, it turns out to be impossible to simply impl <T:Hasher> digest::Update for T in multihash-derive because of rustc --explain E0210)

[hanabi1224]

(The original 3-space indent seems to be inconsistent with the rest of the crate)

[vmx]

Thanks for trying different things.

Also your own comment on the whitespace was really useful. Overall, great work!

[vmx]

@hanabi1224 Do you plan any other changes atm? I'm asking as I would release a multihash-codetable v0.2. But if there are further changes soon, I'd wait for that.

[hanabi1224]

Hey @vmx Thanks for the quick review! I don't have more changes to propose atm.

[vmx]

The 0.2 release is out!