Documentation for plausibleBeforeDeath and plausibleStartBeforeEnd

[MaximMoinat]

Fixes #541. Needs to be rendered with pkgdown after review.

[MaximMoinat]

@katy-sadowski, some thoughts while writing the docs:
For plausibleStartBeforeEnd, I am doubting whether it is CDM Convention or Critical. I argue the latter, as end dates before start dates creates negative durations which might break analyses. This would also mean we should lower the threshold to 0%, always failing if any records affected.

For plausibleBeforeDeath, we have a 60 days grace period. I've explained it as that administrative events are expected until the next month. Or was this an arbitrary choice.

[katy-sadowski]

Woohooo thanks so much Maxim!

Re: the 60 day buffer - I found the explanation on the Themis docs: https://ohdsi.github.io/Themis/obs_periods_for_ehr.html#observation-period-end-date. I do wonder if this warrants its own general convention for "How to handle events dating after death"? @clairblacketer what do you think? [PS - super exciting to be able to search for and find that on the website 😄 ]

Re: start before end - interesting question. I am torn, because this could cause really serious problems in some cases, maybe even errors in code, so that would be an argument for Fatal. On the other hand I think the intention of Fatal was really database conformance level stuff. @clairblacketer what do you think?

[MelaniePhilofsky]

We do have a ratified convention for events after death. It needs to be documented in the Themis convention library. Right now it is just a ratified and closed in the Themis issues OHDSI/Themis#42
I will reopen it as a "needs documentation"

[katy-sadowski]

Thank you @MaximMoinat !

Maybe we can discuss Fatal vs Convention for StartBeforeEnd at the next DQ call. Can merge in for now and update if we decide to change it later :)