This PR introduces the PixelMap class for representation of a 2D array of pixel values. It includes math functions to add/subtract/divide/multiply with scalars, arrays or other PixelMaps. It includes a thresholding feature to set values below (or above) a threshold to np.nan. It includes an imshow method to make plotting the pixel map easy.

Subclasses set default values for plotting. There is a convert_to method to convert to different subclasses.

I defined a set of subclasses that can be expanded later.

• TIVMap for TIV or amplitude maps
• ODCLMap for overdistention/collapse maps
• PerfusionMap for perfusion data
• PendelluftMap for pendelluft data
• SignedPendelluftMap for pendelluft data that include negative values indicating the direction of phase change
• DifferenceMap for differences between other maps

I've written this according to some new 'best practice' rules:

• At initialization, time (not the case here) and values can be supplied as positional arguments, the rest can only be supplied as keyword arguments.
• All classes holding data and their attributes are frozen, i.e. immutable
• Classes can be updated using an update method that returns a copy with the keyword arguments replacing attributes.

I included a test_pixelmap_imshow notebook with examples and tests for the visual part of imshow.

Included in this PR is a slight restructuring of the documentation. I can extract it to a different PR if necessary.

I've tried to keep with default colormaps, or use black -> color. However, I think these can be tweaked a bit further, e.g. with black -> color -> white. Any suggestions are welcome!