Performance: use nom_parser in lopdf instead of pom_parser

[badicsalex]

Apparently lopdf is also changing to nom_parser as a default, but since this is forced in pdf-extract's Cargo.toml, it should be modified there too.
The performance change is dramatic, for one of my sample files I get a 60%+ reduction in runtime.

See J-F-Liu/lopdf#157