Thanks for this awesome SVD/PCA implementation!

I've made a fair bit of changes that hopefully improve the code. I've structured the commits such that it should be fairly simple to exclude anything if needed and it will be easiest to review this PR commit by commit.

I've done the following:

1. Tests are generally kept in a separate file or directory. This makes everything much more readable. I guessed I probably need to include the copyright in the test file as well.
2. Black is a fairly standard code formatter that makes the code more readable.
3. There were a lot of empty comment lines which made everything much harder to read, so I removed those.
4. The reason I even started this - add the random_state parameter to all the functions. check_random_state was taken from scikit-learn to be consistent (adding scikit-learn as a requirement would be overkill for the simple function). Requiring the user to set a global random seed is not very user-friendly and this method is very familliar in python.

If you don't like any of formatting changes or moving tests to their own file, I can remove that, but having a random_state parameter would be really useful.