Describing the spread of outcomes from simulations In general, what is the term to describe the 5th and 95th percentiles of outcomes coming from a simulation study?
Say we obtain a vector of outcomes y. In R we can obtain the 5th and 95th percentiles as: quantile(y, probs=c(0.025, 0.975))
Would this interval be called the 95% confidence interval, 95% credible interval, 95% percentile interval, 95% quantile?
In my case I am using an agent-based model where the input parameter values for simulation are sampled from a posterior distribution, in case that makes a difference. 
 A: Maybe it is better to paraphrase than to look for a precise technical term. First, if you don't know the term, chances are your readers don't know either. Second, using less nouns is considered good style.
So one could write:


*

*90% of the observations lay between $a$ and $b$.


This is not completely precise.  But it is as precise as talking about confidence intervals without specifying the underlying method (or whether they are symmetric). An alternative is:


*

*5% of the observations lay below $a$, and another 5% lay above $b$.

A: It's the interval between the 5th and 95th percentiles. That should be easy to explain. 
Of the alternatives you mention: 
confidence interval might be satisfactory or completely wrong, depending on what you are simulating. (You do explain, but I don't understand enough of your explanation to advise either way.) 
95% quantile is wrong, because that can only plausibly mean the same as 95% percentile. There is a convention that quantiles are characterised by fractions between 0 and 1 and percentiles by percents between 0 and 100, but that's just a convention and those fractions and percents are equivalent in any case. I don't think there would be support for using that term for an interval 
credible interval means what you want it to mean, I guess, but I don't think it's standard here or a good term of art in any case. Bayesians (I am not antagonistic, just not fluent in their terminology) might want to comment further. 
95% percentile interval -- if unexplained -- I might guess as the interval between 2.5% percentile and 97.5% percentile, just as 95% confidence interval conveys (intended) coverage, so that is not a good idea. 
A loose analogy is with plots in statistical graphics. Many researchers, and even some authors on the subject, seem to think that every distinct kind of plot should have a distinct short and punchy name, but that is not  essential, or even necessarily a good idea. Similarly with intervals: it's more important to be clear about what the interval covers than to use a punchy name. Giving some detail in a definition just once is unlikely to irritate anyone and is helpful to all. 
