When doing count_over_time and rate(), it doesn't make sense to have a line_format since you're counting line, you don't really care about what the line contains.

At the same time this set the foundation of for further optimization, like parsing only required labels during label parsing.

I've also found a bug where `topk by(foo) (1...` where not correctly parsed.

Signed-off-by: Cyril Tovena <cyril.tovena@gmail.com>

Except for bytes one which count bytes in lines.

Signed-off-by: Cyril Tovena <cyril.tovena@gmail.com>

Co-authored-by: Owen Diehl <ow.diehl@gmail.com>