Rearrange the benchmark code so that each implementation is
self-contained. Adds the SierpinskiTriangle case that 'emotion'
introduced in their fork of the 'react-native-web' benchmarks. And make
it possible to run benchmarks on a per-library basis.
Introduces a monorepo structure, relies on yarn workspaces to share
dependencies, and lerna for syncing versions across the monorepo.
* Create 2 workspaces:
'packages' and 'website'
* Create 2 public packages:
'babel-plugin-react-native-web' and 'react-native-web'
* Create 1 private package:
'benchmarks'
A simple release script runs the tests, builds the package assets,
increments the package version numbers, git commits and tags, publishes
the package to npm, pushes the changes to github, and releases the
website update.
Close#657
Nicolas Gallagher