PGN parsing

[PenguinF]

Following these specifications.

• [x] Develop parallel to or in analogue with Json parsing
• [ ] Research use of System.Memory.dll (ReadOnlySpan...)
• [x] Use red-green syntax nodes, which helps with reducing memory allocations while editing+live-parsing PGN
• [ ] Support several encodings in the import format. Allow Unicode in comments, and arguably strings.
• [x] NAG support.
• [ ] Make some decision about how the multi-line comment delimiter character } can be represented without delimiting the comment. It seems that the PGN standard doesn't provide any kind of escape mechanism. For interoperability with existing PGN software it makes sense to research how such software resolves this.
• [ ] Separate action for strict exports, in which the level of strictness can be tightly controlled: max line length, strict line endings or not, encoding (strict ASCII? UTF-8? allow characters from the discouraged 0xa0-0xbf range? etcetera)
• [x] Allow angle bracket characters < and > ? 'Reserved for future use' but then what is the expected behavior?
• [x] Add appropriate benchmarks

Use a state-machine-with-stacks parser rather than anything recursive like in the Json parser, since this is vulnerable to StackOverFlowExceptions. This also has the advantage of more naturally supporting FileReader vs. existing string scenarios; the latter can be accessed with Span, the former with sequences of Spans while leaving the parser in a valid intermediate state. And this has the potential of allowing reconstruction of a PgnParser state given a red syntax tree which helps with live parsing of edits.

[PenguinF]

Angle brackets just not allowed. Will generate an illegal character error.