upgrade to Crypto++ v5.6.3

[zooko]

Crypto++ is under new maintainership, led by Jeffrey Walton. I'm one of the co-maintainers who participates in the private email discussion about what changes to make and so forth. IMO Jeffrey Walton is doing a great job of quality assurance, testing, careful implementation, code-auditing, etc., and I feel very good about increased safety/reliability of Crypto++ under his watch. Here's his release announcement for Crypto++ 5.6.3:

​http://www.mail-archive.com/cryptopp-users@googlegroups.com/msg08386.html

I think pycryptopp should upgrade to Crypto++ 5.6.3 ASAP (although I *also* still think that we should disable assembly optimizations ASAP! — #85 — most of the bugs and compatibility issues that Jeffrey has been fighting would disappear with that change!)

[daira]

+1. Normally I would recommand OTAAT, but in this case I think we should do both at the same time, so that we don't have to worry about possible regressions in the assembly code for Crypto++ 5.6.3. (Even if we don't think there are likely to be any such regressions, it is better just to exclude the possibility.)

[dawuud]

i got my dev branch to build with cryptopp 5.6.3 here:
​https://github.com/tahoe-lafs/pycryptopp/pull/29

however i think we should instead use their latest because this would also help to solve for adding two more crypto primitives: blake2 and chacha20.

i've commented on the progress here on the chacha20 ticket:
​https://tahoe-lafs.org/trac/pycryptopp/ticket/94#comment:7

[dawuud]

moving this to ​https://tahoe-lafs.org/trac/pycryptopp/ticket/100
where we will upgrade to the latest stable