test: add unit tests for lexer raw string literals, lone dash, and single equals

[Aiudadadadf]

est_lexer.py had no unit-level coverage for three distinct code paths in lexer.py, even though the compliance suite exercises some of them indirectly at a higher level.

The first gap is the raw string literal path (the ' branch in okenize). _consume_raw_string_literal handles basic content, preserves interior whitespace, and applies
eplace("\'", "'") to unescape embedded single quotes, while leaving other backslash sequences intact. None of those behaviors had a dedicated unit test; four tests now cover them directly.

The second gap is error handling for two invalid token forms. A bare - with no following digits reaches a len(buff) > 1 guard that raises LexerError, and a = not followed by another = raises LexerError via a branch with distinct EOF vs. non-EOF position logic. Three tests now exercise these paths, including the EOF edge case. All 999 existing tests continue to pass.