Add etas to results table

[mccarthy-m-g]

This PR adds etas to the results table. I opted to add two types of etas:

• eta<nn> etas, which are the iterative empirical Bayes estimates that pair with iter_ipred.
• map_eta<nn> etas, which are the full-data MAP empirical Bayes estimates that pair with map_ipred. These are the equivalent of the etas reported in a NONMEM output table, and are what you'd use for an empirical eta-distribution plot.

I also updated the documentation for run_eval() to document its outputs more thoroughly, particularly with a @details section describing the results table.

[dominic-irx]

works great if everything works smoothly.

can you write a test to make sure that if all fits fail that the function returns something useful (warning, maybe)? also looks like line 97 (par_dummy) won't have etas.

[mccarthy-m-g]

works great if everything works smoothly.

can you write a test to make sure that if all fits fail that the function returns something useful (warning, maybe)? also looks like line 97 (par_dummy) won't have etas.

Done! c366ba5 addresses the above.

We actually had an appropriate warning already in calculate_stats(), so I refactored that to occur in run_eval() instead: efd88e4. The warning can still be triggered in calculate_stats() when it's run outside run_eval() though