Update crates to digest v0.10

[newpavlov]

Depends on RustCrypto/traits#819.

Notable changes:

• ripemd160, ripemd256, and ripemd320 merged into ripemd.
• Complete rewrite of the groestl crate using (public domain) code from the NIST submission as a reference. Now the crate has an acceptable performance (~100 MB/s vs ~1 MB/s). It can be improved even further by utilizing AES-NI and SIMD in general, but I don't plan to work on it myself.
• Use LE table for Whirlpool, thus improving performance a bit on LE targets.
• Added Tiger2 variant.
• fsb improvements.
• sha2, streebog, and shabal are implemented in terms of the VariableOutputCore trait. For example, Sha512VarCore implements it, but supports initialization only for 4 sizes and its variants are defined as:

pubtypeSha512_224 = CoreWrapper<CtVariableCoreWrapper<Sha512VarCore,U28>>;pubtypeSha512_256 = CoreWrapper<CtVariableCoreWrapper<Sha512VarCore,U32>>;pubtypeSha384 = CoreWrapper<CtVariableCoreWrapper<Sha512VarCore,U48>>;pubtypeSha512 = CoreWrapper<CtVariableCoreWrapper<Sha512VarCore,U64>>;

Closes#67
Closes#317
Closes#330

[tarcieri]

FYI, I just merged Tiger in #229, so you'll probably want to rebase

[atenjin]

Our bitcoin lib need to be compiled in Wasm, 0.9.0 can't work now... I try 0.10 is ok, hoping this pr could be merged soon...
And thanks for your work all in all.

[newpavlov]

0.9 versions should work fine on WASM. If it's a problem with our crates, can you please open an issue with a description and reproduction steps?

[newpavlov]

BTW I wonder if we should synchronize all hasher crate versions to v0.10. It would make it much easier to see at a single glance which trait version a crate implements, but it would mean we will have a discontinuity in minor versions for some crates, e.g. sm3 will jump from v0.3 to v0.10.