Locality Preserving Projection module for manifold learning

[adrinjalali]

scikit-learn/scikit-learn#1628 proposes a new method to sklearn, but it's been stalled and it may be a better idea to include it here. This is for us to decide whether it's a good idea to have it, and if yes, then it shouldn't be too hard.

From the original PR:

---

This pull request adds a new module "Locality Preserving Projection" (LPP) to the manifold learning package. LPP is can be see as a linear approximation to the Laplacian Eigen Mapping. Unlike other manifold learning algorithms, LPP is a linear transformation and can be used like PCA.

Detail of LPP can be found in the following paper.
"X. He and P. Niyogi. Locality preserving projections. Advances in Neural Information Processing Systems 16 (NIPS 2003), 2003. Vancouver, Canada.".

Currently I finished main LPP module and added examples. The remained tasks are to write test codes and documents. Can any one suggest me a guideline to write unit-tests? I don't understand its manner.

Thank you.