Add the Diffusion Map Dimensionality Reduction Technique

The diffusion map is a dimensionality reduction technique that uses a transitional probability as its "distance" measure. It is noise-resistant and non-linear. Moreover, the algorithm itself is fast and scalable (I've used it with 50k+ samples). I wrote a quick implementation here.

It has been of interest in numerous scientific fields:
Diffusion maps (1119 citations)

Geometric diffusions as a tool for harmonic analysis and structure definition of data: Diffusion maps
(672 citations)

Functional diffusion map: a noninvasive MRI biomarker for early stratification of clinical brain tumor response (491 citations)

Diffusion maps and coarse-graining: A unified framework for dimensionality reduction, graph partitioning, and data set parameterization (450 citations)

Furthermore, a Google scholar search for "diffusion map dimensionality reduction" yields more results than "isomap", "locally linear embedding", or "hessian eigenmapping".

If I fix up my implementation, would you all be interested in having this in sklearn?

The issue names no repository files, tests, or entry points. Start by reviewing the linked implementation and the referenced diffusion-map papers; the issue does not define an acceptance test or a concrete definition of done.