Throw warnings for some input variable ranges

[fredjaya]

Such as:

• sensitivity < 0.5
• specificity < 0.5
• $(1-specificity) \geq prevalence$
• correlation = 0 (in effect random sampling)
• correlation = 1

Will need to decide whether these should be in functions, depending on re-usability and need across functions.

[AngusMcLure]

Yes, most of the these should be warnings in most functions. Note that because we have functions calling functions and passing on some of these parameters, you might want to put the warning at just the first called function (or the last called function) to avoid getting the same warning twice (or more) for each function call. E.g. optimise_sN_prevalence works by calling optimise_s_prevalence repeatedly (and will use the same sensitivity, specificity, prevalence, and correlation across all these calls). So we'll need to think a bit carefully about where we put the warning

[fredjaya]

Adhere to the tidyverse style guide for error messages

[AngusMcLure]

Yes, this would be good. I hadn't read these before!

[fredjaya]

Refer to #27 for typical values