Bootstraphistogram

- Add `ignore_nan` option to some methods to use `numpy.nanmean`,
`numpy.nanstd` and `numpy.nanpercentile` instead of `numpy.mean`,
`numpy.std` and `numpy.percentile`.
- Add more flexibility to plotting functions to display percentiles or
mean/standard deviation.
- Improved type hints.

- Improved type hinting.
- Fix linter errors and bugs in CI.
- Add a development Dockerfile.

- Implement BoostrapEfficiency for calculating binned efficiencies.
- Implement binary operation (`+`) on `BootstrapMoment`.
- Improve tests and new documentation.

- Implement binary operations (`+`, `-`, `*`, `/`) on histograms.
- Fix issue [5](https://github.com/davehadley/bootstraphistogram/issues/5).
- Add `BootstrapMoment`, a class that calculates the first three moments of the filled data.
- Improve tests and documentation.
- Speed improvements when filling with small arrays.

- Change BootstrapHistogram API to accept `numpy.ArrayLike` as input.
- Fix issues [1](https://github.com/davehadley/bootstraphistogram/issues/3) and [#4](https://github.com/davehadley/bootstraphistogram/issues/4).
- Improve documentation.