Removes ethereumjs-util and keccak dependencies.

[MicahZoltu]

Replaces both with js-sha3. ethereumjs-util and keccak NPM packages both have native bindings in their dependency tree which means they depend on node-gyp along the way. js-sha3 on the other hand is a JavaScript-native implementation of the sha3 family of hashing functions, which relieves a number of node-gyp and native binding related issues.

Fixes #356
Fixes #336
Part of #361

[MicahZoltu]

Is there anything I can do to expedite this getting merged?

[MicahZoltu]

I believe the coveralls coverage decrease is because I deleted some lines that were previously covered, so the number of covered lines as a percentage went down (uncovered lines have more weight as of this change).

[axic]

Sorry for the delay. I see a reason to not use two packages for keccak256, but what's wrong with ethereumjs-util? It is supposed to be well maintained.

[MicahZoltu]

1. Depending on ethereumjs-utils brings in 30 transitive dependencies.
2. Depending on js-sha3 brings in 0 transitive dependencies.
3. The only function needed is keccak256.
4. A couple of transitive dependencies of ethereumjs-utils depend on node-gyp, which causes no end of problems.

[axic]

I'd prefer depending on keccak as that has a more optimised implementation, but apparently your opinion is it is riddled with node-gyp. Wish the JS ecosystem wasn't this broken.

[coveralls]

Coverage decreased (-0.2%) to 57.267% when pulling f441441 on MicahZoltu:js-sha3 into ea40b79 on ethereum:master.

[MicahZoltu]

Keccack256 is called in two places, nowhere near a hot path from what I can tell. Given the many orders of magnitude slower the compilation process is than hashing, I think choosing the native library for its performance benefits at the cost of increasing dev complexity is very much not worth it.

[axic]

The use in downloadCurrentVersion doesn't matter, that is a publishing level tool.

The use in linker.js matters, because that is a helper to do linking of binaries and is used by Remix, Truffle, etc.