Use postcard to serialize cookie data

[djc]

@cycraig want to review this?

[codecov]

Codecov Report

Patch coverage: 100.00% and no project coverage change.

Comparison is base (fa9156d) 57.32% compared to head (bf81d65) 57.32%.

Additional details and impacted files

@@           Coverage Diff           @@
##             main      #81   +/-   ##
=======================================
  Coverage   57.32%   57.32%           
=======================================
  Files           7        7           
  Lines        1378     1378           
=======================================
  Hits          790      790           
  Misses        588      588           

Files Changed	Coverage Δ
mendes/src/cookies.rs	76.47% <100.00%> (ø)

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

[cycraig]

I don't have a strong opinion on serialization formats but some benchmarks show MessagePack (popular and similarly well-specced) to be smaller for fewer fields, but a bit slower. I only bring it up since the serialization format is a breaking change.

I see postcard does better in benchmarks for strings though. Overall postcard does seem better than bincode in almost every respect (maintenance, specification, size, speed), so should be fine.

[djc]

I think https://github.com/djkoloski/rust_serialization_benchmark is the definitive benchmark right now. From that data, it looks like there's not a big difference on the size axis between MessagePack and postcard. I like that postcard is tightly scoped, works within the serde data model and appears to be well-documented -- I think it will serve us well.