consider adding ensemble details in real-world forecast session #89
Description
This whole section is really cool, but I feel like I'm still wondering (relative to the presentation) how the ensemble is actually constructed. There is a link to "the gory details", but maybe there's a quick way to illustrate it? I feel like there may be a simple illustration somewhere for this? No problem if this doesn't make it in here.
For example, does the basic unweighted LOP take 100 samples at random from all the models for each quantile? What's the default?
There's already a lot here, so maybe this isn't best too.
Originally posted by @trobacker in #65 (comment)