FEAT: tile-interpolated Adaptive Histogram Equalization

[neutrinoceros]

Requires #301
As a continuation of #301, I could add the tile interpolated variant of efficient AHE. Here would be some example valid inputs;

• adaptive_strategy={"kind": "tile-interpolation", "tile-size": (64, 32)}
• adaptive_strategy={"kind": "tile-interpolation", "tile-size": (64, 64)}
• adaptive_strategy={"kind": "tile-interpolation", "tile-size": 64}
• adaptive_strategy={"kind": "tile-interpolation", "tile-into": (8, 16)}
• adaptive_strategy={"kind": "tile-interpolation", "tile-into": (8, 8)}
• adaptive_strategy={"kind": "tile-interpolation", "tile-into": 8}

tile-into would represent the number of tiles to create in each direction, while tile-size would give the size of a single tile as a number of pixels (same as for "kind": "sliding-window").
I would like to de-correlate this as much as possible from boundary conditions support, so as a first step, only "closed" BCs should be added.

Originally posted by @neutrinoceros in #290 (comment)

Linked PRs:

• ENH: implement tile-interpolation AHE strategy interface #365
• ENH: implement tile-interpolation resolution #371
• ENH: connect tile interpolation AHE backend and frontend #375
• ENH: tile interpolation AHE backend #377