rfc3339 direct ser/deser fix for protobuf Timestamp

[greg-szabo]

A fix for an issue that bubbled up during #665 .

The Time domain type was incorrectly serialized by itself. (It was correctly serialized as part of another struct.)

This change implements correct serialization for tendermint::Time via tendermint_proto::Timestamp.
Timestamp needed a helper struct so it could be serialized/deserialized without custom serializers.
You can clearly see the dependencies on serializers in this change:

• The serializer to correctly ser/deser an RFC3339 string into a Timestamp is moved into tendermint_proto::serializers::timestamp.

• The Timestamp struct needs a helper struct so it can use those serializers. Otherwise Timestamp would have required a custom serializer.

• With the Timestamp struct directly serializable, there's no need for special serialization for Option<Timestamp>. All structs having that field can now be serialized by just defining serialization over Option<> with the tendermint_proto::serializers::optional module.

• Unfortunately, the tendermint::Time struct is directly serialized/deserialized in testgen structs (and possibly in IBC) in several places. But now that it is directly serializable through Timestamp (and from an RFC3339 string), there is no need for extra annotation when serializing/deserializing it. It's going to be interesting if/when someone comes to us that they need it serialized differently (in the {seconds:x,nanos:x} format, isntead of RFC3339).

• Referenced an issue explaining the need for the change

• Updated all relevant documentation in docs

• Updated all code comments where relevant

• Wrote tests

• Updated CHANGELOG.md

[romac]

I like the new Rfc3339 struct :)

Just suggested a whitespace change in the README, but otherwise LGTM :)

[andrey-kuprianov]

@greg-szabo What I especially like about this PR is that it removes more code than it adds -- it simplifies things instead of complicating them 💯