Powerbox

------
**Bugfixes**

- Fixed cubegrid return value for dft functions when input boxes have different sizes on each dimension.

------
**Features**

- Added fft/ifft wrappers which consistently return fourier transforms with arbitrary Fourier conventions.
- Boxes now may be composed with arbitrary Fourier conventions.
- Documentation!

**Enhancements**

- New test to compare LogNormalPowerBox with standard PowerBox.
- New project structure to make for easier location of functions.
- Code quality improvements
- New tests, better coverage.

**Bugfixes**

- Fixed incorrect boxsize for an odd number of cells
- Ensure mean density is correct in LogNormalPowerBox

------
**Bugfixes**

- Fixed bug in pyFFTW cache setting

------
**Enhancements**

- New interface with pyFFTW to make fourier transforms ~twice as fast. No difference to the API.

------
**Features**

- New functionality in ``get_power`` function to measure power-spectra of discrete samples.

**Enhancements**

- Added option to not store discrete positions in class (just return them)
- ``get_power`` now more streamlined and intuitive in its API

--------------------
**Enhancements**

- Improved estimation of power (in ``get_power``) for lowest k bin.