Add option to classify components based on their correlation with regressors

[eurunuela]

Summary

Stems off #879. A number of users have mentioned in the past they would love tedana to consider regressors (good and bad) to better classify ICA components.

Additional Detail

With the new decision tree, it should be approachable to add a node that takes a regressor file with some other arguments and compares them with the ICA component time series. Then based on a threshold applied to correlation values, tedana could classify components as accepted or rejected. This could also be extended to maps in the future.