@iarna/toml

This release features an overall 15% speed boost on our benchmarks. This
came from a few things:

• Date parsing was rewritten to not use regexps, resulting in a huge speed increase.

• Strings of all kinds and bare keywords now use tight loops to collect characters when this will help.

• Regexps in general were mostly removed. This didn't result in a speed
  change, but it did allow refactoring the parser to be a lot easier to
  follow.

• The internal state tracking now uses a class and is constructed with a
  fixed set of properties, allowing v8's optimizer to be more effective.

In the land of new features:

• Errors in the syntax of your TOML will now have the fromTOML property
  set to true. This is in addition to the line, col and pos
  properties they already have.

  The main use of this is to make it possible to distinguish between errors
  in the TOML and errors in the parser code itself. This is of particular utility
  when testing parse errors.