Added draft implementation of factor analysis

[jfsantos]

This is a draft implementation of factor analysis based on the algorithm described on Bayesian Reasoning and Machine Learning (David Barber) and the scikit-learn implementation. I still did not optimize it for performance so it may have some issues (e.g., allocating memory for matrices inside the loop). I also did not write any tests but I am planning to compare it to the scikit-learn implementation on the same dataset.

Is there interest in adding it to this package if I manage to improve the implementation and write tests and documentation? Any suggestions to improve the implementation or coding style would be appreciated.

[lindahua]

Factor analysis is definitely within the scope of this package.

Would you please add some tests to ensure the correctness?

[jfsantos]

I managed to get the fit function performing similarly to the implementation in scikit-learn. I still have to fix the transform implementation and finish writing the tests. After that I will add some functions to compute scores (not strictly necessary if only using the factor analysis model to perform dimensionality reduction, but I think it would be a good tool).

[wildart]

Superseded by #33