minor fixes

[ceteri]

• fixed minor typos and a few broken links

---

Misc. notes:

• great use of the repr() method for the SimpleContextualBandit environment. I'll start using that as well.
• I like the exercises in the early part, to perturb rewards then see the impact.
• from ray.tune.progress_reporter import JupyterNotebookReporter -- good to see this
• In the "solutions" notebook, it's so interesting to see the comparison of LinTS vs. LinUCB given the non-linear nature of the problem domain.

FWIW, when running in JupyterLab the (image) link in 04-Linear-Upper-Confidence-Bound.ipynb doesn't work (i.e., hyperlinking to a file in a parent directory?) That may be a matter of JupyterLab security. I find there are different semantics for rendering notebooks in JupyterLab vs. GitHub, and your selected trade-off is good. In any case, the Boken plot for it runs fine so that covers the JupyterLab side.

NB: in the 06-Market-Example.ipynb notebook the MarketBandit.seed() never gets used in our example -- although it's probably better to leave it in. For example, a user might run a more extended example and use that to analyze the bandit behaviors.

[ceteri]

Version diffs shown in GitHub for 06-Market-Example.ipynb have a bunch of minutiae included from the Boken run and perhaps also my version of JupyterLab rendering differently (e.g., different CDNs)

The main change that I made was toward the end, a minor rewording of the paragraph A final note: when you checkpoint...