Install empiricaldist from conda-forge

tl;dr Install empiricaldist with conda now that it is available from conda-forge

Background

A minimum of 0.6.7 is required because version 0.6.6 is broken (it doesn't import matplotlib, which it requires for its plotting methods)

I confirmed that I could build all the main chapter notebooks with this updated environment:

mamba env create --file environment.yml
conda activate ThinkBayes2

for nb in chap*.ipynb
do
  jupyter execute $nb
done

I also attempted to build the non-chapter notebook, but clustering.ipynb fails because the environment doesn't have scikit-learn installed:

ModuleNotFoundError: No module named 'sklearn'

Given how it already takes so long to solve and build the conda environment, and also that the clustering notebook is not included in the online version of the book, I decided to not add it in this PR.

Also note that per our discussion in https://github.com/AllenDowney/ThinkBayes2/pull/56#issuecomment-1117536918, I took a stab at creating two separate environments: one for readers, and one for you to build the book. I loss motivation for now because:

The reader-only env wasn't that light-weight anyways. The final few chapters ended up requiring a bunch of packages that I had hoped to remove
I ran into the frustrating pymc3/arviz error that you had previously run into. Pinning to old versions of these dependencies limits the long-term usefulness of the book's code (eg it can't benefit from bug fixes in later releases). When I tried to force conda to install newer versions on macOS, the conflict was due to some low-level package (I can't remember which), which would be non-trivial to fix or workaround
I got errors like missing scikit-learn in clustering.ipynb (as mentioned above). It's unclear to me which code is expected to be able to be executed using the conda env

In case it could be useful to you, here's what I had when I stopped: