Scrape-schema-recipe

* fix to allow `type` to be a List.
* add `py.typed` file
* minor fixes

all from PR 32

* Fix output to remove HTML character references such as `&` and replace with the proper Unicode symbol (so `&`)
* Improve typing.
* remove keys with empty values (`None`, `[]`, or `""`) from the outputted recipe.

Changes made in PR 30

This is from PR 27.

Created custom `SSRTypeError` (one item of Issue 26), this is technically an API change. For most it won't matter too much. If you were catching a `TypeError`, now catch the custom `SSRTypeError`.

For Python 3.6 `dataclasses` backport is required for the SSRTypeError and future changes.

For the file `test_scraper.py`:
* corrected a few typos
* converted from CRLF (Windows/DOS) to LF (Linux)
* because of those extensive I just went ahead and converted the code using black code formatter.

This is a small step on the road of 1.0.0.

* Unset values for date values are removed from the resulting recipe when using `python_using` is set to `True`. bugfix (PR 24)
* Add an example from MedlinePlus

PR 20 -
* Remove validators library (fixing Issue 18)
* Add another example.

fixes `example_output()` so that it works (Closing Issue 15)