* added dev-venv target

* added subextract model

* added nltk dependency

* flake8 ignore line break before binary operator

* moved dev dependencies up

* added nltk punkt download

* added nltk download to dev-venv; pytest and pytest-not-slow target

* added subextract training pipeline

* added optional xpath namespaces

* log failed xml file

* use recover parser option

* added subextract app

* start subextract server

* renamed to autocut

* declare slow and very_slow pytest markers

* make autocut main test as slow

* fixed post data

* updated README

* also build non-dev image as part of ci

* added pytest.ini to dev image