Lexer written using the logos with tweaks/benefits:

• A lot of complexity is abstracted thanks to logos.
• All of tokens are now lexed properly incl. opcodes, hex/string literals etc. Able to lex entire erc20.huff without errors.
• Opcodes/evm types based on context e.g. address is both a evm primitive type and opcode, but it'll be lexed as opcode only when used within a macro body, as type only when inside param list of functions/events and as ident otherwise. Same goes with EVM primitive types and array types.
• Eof and Whitespace are removed. Whitespaces are simply ignored and EOF can be simply inferred when lexer iterator gives out None value.
• Lexer's iterator's assoc. type is now simply Token instead of a Result. So, you just do token.unwrap() instead of token.unwrap().unwrap().
• A notable change is that TokenKind enum variants hold data (if any) of type &'a str only avoiding mixed types - making it more memory efficient because of the way enums behave in rust. Necessary conversion is done at parser or wherever required.

lmk any suggestions :)