Accept a value to ignore covariates when passed to differential_expression

[kelshmo]

stepwise_regression selects a model and passes it to generate top ranked differential expression gene lists.

There are cases where covariates are not independent and therefore, need to be excluded from the formula for a particular contrast.

An example use case: Diagnosis is not independent of, or is orthogonal to, Braak and CERAD scores. When diagnosis is the response variable, the model should not correct for orthogonal scores. Conversely, if Braak is the response, you might want to correct for diagnosis.

[kelshmo]

Another example use case: I want the complete model documented in the report, but I want to manually exclude some variables from diffferential_expression().