Name To Structure: implement suffix-based lexeme stream analysis

Currently, a lexeme stream, that we get after tokenization phase, is parsed in left-to-right order with some forward parsing where applicable. While this approach works for simple names and structures, this will become cumbersome and error-prone in the future. The idea is to make parsing suffix-based, i.e. to split lexeme stream in several parts based on suffixes and perform further processing based on those smaller parts. This approach might significantly lighten overall parser codebase and make parser more maintainable and extendable.