This includes:
- ensuring we don't go off the end of input if it's not null-terminated
- checking for overflow and underflow
- being more careful to check that numbers match JSON lexical syntax

This is the first step towards JSON decoding, implementing the basic
functionality of JsonCodec::decodeRaw. The main outstanding issues are:

- it allows trailing commas in arrays and objects
- it is too liberal in number syntax, eg allowing a leading +
- it does rejects non-ASCII characters in \u escapes

Refs https://github.com/sandstorm-io/capnproto/issues/255

Fix some documentation typos.

Fix typo in documentation

Add TextCodec to expose an API for the schema's text format.

Spellcheck the styleguide

Typo found by @mquandalle

Mentioned three days ago, didn't want it to get forgotten

Fix typo in two party RPC docs

Make sure ArrayPtr<char> stringifies the same as ArrayPtr<const char> (without commas after each character)

Use ip4 sockets for wildcard on OpenBSD