GitLab

Refactored the operator >> and Node::Read default functions, as well as the conversion functions, to more easily read new types as keys (this uncovered an error, in example 2.11 of the spec)

Added templated casting to nodes, as well as operator == and != (for quick checks, especially to help in testing). Implemented size() on a map node to return the number of key/value pairs (as in std::map)

Fixed bug in anchors with no content. This involved refactoring the 'implicit sequence' concept (where a map and a sequence start on the same indent, but we read the sequence as more indented since the '-' is visually an indent).

Switched from loading test files to testing specific parsing constructs. The tests don't fully cover the span (eventually I'll add more, maybe), but there's a bunch there.
More to the point, the yaml-reader program doesn't do any file IO, so it doesn't require a specific working directory.

It's better organized now, I think - nodes only offer a single main way of getting the fundamental scalar (as a string), and now we can specialize a single template to read specific types.

In particular:
1. Windows CR/LF weren't read properly (issue #11)
2. Scanning wasn't reading EOF properly
3. Documents may be empty (this was old, I think)
Also fixed some VS2008 warnings on /W4.

Switched to reading the entire file into a buffer at the start.\nThis speeds it up a TON (like 100x).

Added a templated derived exception to KeyNotFound so that you can figure out *which* key wasn't found.

(Thanks to Vadim Zeitlin)