This is a no-op change that standardizes the imports in sql.py, which makes the file easier to work with for future improvements.

Clean up sql.py

Added a minimal sample that takes 20 seconds to parse as `regression-2`.

Removing the `_space` from the comment regex (`_s`) makes this go
instantly, as does replacing `re` with `regex` (replaced it in
lexer.py)

Use negative lookahead to avoid runway backtracking in whitespace.

Fixes #2839

fix lua regex causing runaway backtracking.

Need to disable the format string rule as we're failing that, a lot.