Add additional donor normalization options

[rmarkello]

The issue

The current method for normalizing between-donor microarray expression differences prior to collapsing across donors is to use a scaled robust sigmoid correction procedure. This is cool and good and I think the right move, but is perhaps a bit non-standard so providing alternative options would be nice.

Proposed solution

Adding a parameter to abagen.get_expression_data() like donor_norm and adding the ability to z-score the data would be cool. That way someone could run either:

abagen.get_expression_data(atlas, donor_norm='srs')

or

abagen.get_expression_data(atlas, donor_norm='zscore')

and get back interpretable results.

My only concern with this solution is that the number of optional parameters for abagen.get_expression_data() is getting a bit overwhelming! It might be worthwhile making this function a "workflow" (with a CLI, too?) and refactoring some of the other functions to be more modular / tool-oriented so that a user could construct an alternative workflow if they wanted.