Parts of speech cleaning

[jpcompartir]

Add to LimpiaR to form part of pre-processing - plus dances nicer with the dependencies as LimpiaR is v lightweight still

[jpcompartir]

Heavily using {udpipe}

First wave:

• load model
• annotate data frame in pipe-able way
• re-create sentences/documents from tokens/lemma etc.
• parallel processing via udpipe built in
• progress updates via udpipe built in

Second wave:

• integration with other company packages
• visualisations using output (probably sit in a different package for deps.)
• abstractions for the most common steps e.g. swap only verbs for their lemma and repair sentence/document + join back to original data
• noun-phrase extraction + 'consolidation' e.g. similar to spacyr

[jpcompartir]

Underway in #63