Histbook

Profile plots have been tested in PySpark.

Fix bug in Spark filling.

Added SparkSQL filling: `hist.fill(df)` in PySpark uses Spark to compute increases to histogram bins and then adds them into `hist`.

Tested: all axis types, weights, and some simple expressions.

Implemented but untested: profile plots and `Book.fill`.

Note: no thread safety yet (not just for Spark, but in general).

Reduced the pickle size of `Hist` for easier remote filling.

Plotable` has first-class methods for `.ipyvega()`, `.altair()`, `.vegascope()`.

Now with a [Binder tutorial](https://mybinder.org/v2/gh/diana-hep/histbook/master?filepath=binder%2Ftutorial.ipynb).