Retrain w/ surrogate data

[jstout211]

• Add varying noise levels to temporal sequences
• Resample temporal sequences (290 - 310 Hz) (?)
• Project to SSS operators and rotate + translate to new topo
• Add some fuzz to topo
• Multiply topo & time series by -1 (this is an equivalent solution)