Few improvements to Logistics example

1. No need for HD polygon here actually. You're not using it except for plotting. Think you can leave it out.
2. The first locations list:
- Only give the addresses and demands as assumption and use `geocoding` (the Pelias endpoint pls) to get the coordinates
- change the order to `address, demand, long, lat`
3. I generally like the approach with `numpy` and `pandas`, but try to keep your variables named, i.e. try to avoid to do e.g. `route[:, ::-1].T`. It's best done in dictionaries, I find, where you can put named keys. It makes it a lot nicer and easier to read.
4. `reversed` for an iterator is preferrable over `[::-1]`. Both for performance and readability.

Apart from that, very well done. I know, it can be hard, but try to think most practical with these examples, a fluid storytelling with just enough info to solve a certain problem in the most elegant and readable way.