Dynamic skill extraction + minor refactoring

[joriscram]

In my task to review 440 graduate resumes I came across this repo and slightly adjusted this great initiative, some of it might be useful for others hence this pull request:

• Main functionality of this pull request is that one can add extraction fields in the config and the app will dynamically extract them from the resumes with string matching. (so no coding involved)

in addition:

• Removed spacy dependencies. NLP models can be imported directly. So you do not need the assets folder and installation (especially on windows machines) is more convenient without spacy.
• My dataset contained anonymous resumes, so had to escape for this situation in the name extraction
• extract function does not need nlp model. In addition utf-8 encoding is ensured by a function and not by forcing the system default encodings.

[bjherger]

@joriscram : PR review provided. Please respond to the feedback.

After one iteration, we should be able to merge this in. Thanks for your input!

[bjherger]

@joriscram Thanks for this awesome contribution. I've validated that the new functionality and spacy asset loading works, and I'm happy to merge this in.

[bjherger]

+1, LGTM. Merging in w/ +1s from myself and @joriscram. No open requests or comments on PR.