Remove extra space in mcmc_areas

[jgabry]

Fixes #218

Copying from my comment in #218:

after playing around with a bunch of options and trying with different numbers of parameters, what ended up working best (I think) was to use an additive expansion that decreases with the number of parameters. I put this on a branch remove-extra-space-mcmc_areas. Here's a comparison with the current master branch:

For 2, 4, 10, and 20 parameters (simulated from standard normals) here is what the current master branch produces:

and here is what the version in this PR looks like for the same parameters:

[codecov-commenter]

Codecov Report

Merging #230 into master will increase coverage by 0.00%.
The diff coverage is 100.00%.

@@           Coverage Diff           @@
##           master     #230   +/-   ##
=======================================
  Coverage   99.28%   99.28%           
=======================================
  Files          31       31           
  Lines        3928     3939   +11     
=======================================
+ Hits         3900     3911   +11     
  Misses         28       28           

Impacted Files	Coverage Δ
R/mcmc-intervals.R	99.22% <100.00%> (+0.02%)	⬆️

---

Continue to review full report at Codecov.

Legend - Click here to learn more
Δ = absolute <relative> (impact), ø = not affected, ? = missing data
Powered by Codecov. Last update f4ed652...ae68e50. Read the comment docs.

[tjmahr]

While we are under the hood for this one, I'd like to fix the clipping where the line in the top panel gets cut off.

The demos you supply here appear to fix this problem, but I can still get it if I use area_method = "equal areas" in this version.

I'm trying the geom_blank() strategies of adjusting plotting area by plotting invisible data.

[tjmahr]

Can you try out your diagnostic plots using the updated version?

[jgabry]

While we are under the hood for this one, I'd like to fix the clipping where the line in the top panel gets cut off.

Good call, thanks.

Can you try out your diagnostic plots using the updated version?

Yup, will check now.

[jgabry]

Can you try out your diagnostic plots using the updated version?

Do you mean the same set of plots I made above? I made them again (although I forgot to set a seed last time so they'll look a bit different). They still look good after your change. Here's using default arguments:

And here's using prob = 2/3, prob_outer = 0.9, point_est = "mean":