Regenerate models based off of R4 (4.0.1)

[mikix]

Also:

• Fixed the pytest config so that tests can pass again.
• Bumped the fhirclient version to 4.2.0.
• Fixed up some docs.

[mikix]

OK so all the files in fhirclient/models/ are "owned" by fhir-parser. And I changed the templates (see above) to make sure that all the actual models themselves would only see real 4.0.0 -> 4.0.1 changes and not artifacts of changes in the generator.

I've gone through all the models myself and confirmed that you only see:

• Comment changes like this
• Comment changes on some properties (usually enums) that list more values than previously.

The test files however, I didn't find as easy a way to avoid churn in. But it was less important, so I didn't try very hard. Sorry. But hopefully those don't need close review.

We're still not running the tests during any CI - I hope to fix that later, but wanted to get on a known-good model-generation baseline first. The old generated tests were wonky.

[mikix]

This was so old that I figured I'd just delete it and start over fresh with a new CI test approach later.

I did confirm that pytest passes if you manually run it and provide the downloaded example files.

[dogversioning]

I'm taking your word at face value on the generated models - if you want me to side by side them against the FHIR spec, lemme know.

[mikix]

I'm taking your word at face value on the generated models - if you want me to side by side them against the FHIR spec, lemme know.

Yes I'm confident. I just went through them again with git diff -- fhirclient/models ':(exclude)*_tests.py'.

HOWEVER, I did forget to mention an actual change. This is in a few classes. A few times in the spec, a class will have a "choice" field that can be different types (like valueDateTime and valueBoolean). And sometimes that choice is "any element type" -- see the defaultValue[x] and fixed[x] fields in ElementDefinition.

In 4.0.1, the element type "Meta" got added to all those choice fields. So if you look at elementdefinition.py in this PR, you'll see an example of that. It's all just additive. And it does seem to be correct per the spec.