Default to r_eff=FALSE for loo() method

[jgabry]

Submission Checklist

• Run unit tests
• Declare copyright holder and agree to license (see below)

Summary

Changes default from TRUE to FALSE for the r_eff argument to the loo() method. This is based on @avehtari's comment at https://discourse.mc-stan.org/t/loo-subsample-slower-than-loo/39556/5

Copyright and Licensing

Please list the copyright holder for the work you are submitting
(this will be you or your assignee, such as a university or company):
Columbia University

By submitting this pull request, the copyright holder is agreeing to
license the submitted work under the following licenses:

• Code: BSD 3-clause (https://opensource.org/licenses/BSD-3-Clause)
• Documentation: CC-BY 4.0 (https://creativecommons.org/licenses/by/4.0/)

[jgabry]

@avehtari There's not much to review here. I requested your review just to confirm that you definitely want to do this and to make sure the documentation is ok. The code change is just TRUE -> FALSE.

[avehtari]

I added one suggestion.

Usually, people check reasonable convergence and ESS before running loo() anyway, so in most cases making it FALSE should be safe, and we can add a comment about this somewhere. In the early days I assumed that MCSE might be bigger, but it most often is small if the sample size is reasonable and all k<0.7.

[jgabry]

This is passing everything now, but wanted to check with @andrjohns before merging this: if we're changing a default argument would you prefer that I add a warning or message if the user calls the loo method without changing the new default? On the one hand, I guess this could be considered a breaking change (although wouldn't lead to any errors it changes default behavior). On the other hand, adding a warning/message would result in a warning or message most of the times loo() is called, which is kind of annoying. And this is a pretty small change and it's documented and we can put it in the release notes. I'm a bit torn on whether we should add a warning/message. @andrjohns any preference?