[nitely]

Fixes #37

Changes:

• Deprecate cbor_serialization/cddl/parser
• Add cbor_serialization/cddl/parser2
• Add CddlNode node based AST
• Add proc parseCddl*(source: string, ignoreComments = false): CddlSchema, convert a CDDL into an AST
• Add proc toCddl*(n: CddlNode, pretty = false, verify = true): string which converts an AST to CDDL. It also parses the resulting CDDL and checks the AST is the same as the original.
• Add proc dumpTree*(n: CddlNode): string, similar to Nim macro dumpTree, it returns the AST structure.
• Add func isValid*(n: CddlNode): bool =, checks an AST structure is valid.
• Add roundtrip tests. The toCddl cannot keep the original format, but it should produce the same CDDL schema; so the roundtrip test is done on a pre-formatted schema that matches toCddl output.
• Add cbor_serialization/cddl/parser2_schema.md which contains a cddl of the parser Node. It can be used to generate a matching DS. The type_generator is not able to parse it yet (requires union support).

[nitely]

@arnetheduck is cbor_serialization/cddl/parser2_schema.md what you meant by schema of the schema? it can be used to encode a schema into CBOR for example.

For the toCddl (printer) output, you can check tests/test_cddl_parser_36_dump_pretty.txt to see what it produces. It also checks the resulting AST matches the original one (similar to what nph does I think); so it cannot construct a different/invalid CDDL; there is a param to disable it because it can be slow (it needs to parse the output).

No need to review the code I think. It's mostly tests, though. One thing to notice is the CddlNode has the =copy disallowed in Nim +2.2 with ORC to avoid unintended AST copies; an alternative would be making the CddlNode a ref, but then you cannot store it in a const (although, it's still not clear to me what's the use case).