Convergence errors could be described

[tgerke]

If we know this is a convergence problem, let's write an informative error that tells the user this is the case.

> riskratio(
+   formula = death ~ stage + receptor, 
+   data = breastcancer, 
+   approach = "glm"
+ )
Error: no valid set of coefficients has been found: please supply starting values

This is relevant to models.Rmd

[stopsack]

Great idea! Indeed, trying to fit a plain vanilla log-binomial model (approach = "glm") on the breastcancer dataset is the motivating example for risks and prior papers like the one by Spiegelman and Hertzmark.