possible change: remove thresholding from brain objects

Instead of having kurtosis thresholding built into brain objects, we could write a separate function that takes a brain object, filter and threshold and returns a new filtered brain object. this would make brain objects more generalizable outside of the supereeg context and simplify the brain object code. the downside is that we'd have to add an extra line of code, and make a copy of the data, which will be problematic for (very) large datasets

The issue names no files, tests, or entry points. First locate the brain object implementation and its kurtosis thresholding behavior, then clarify the separate function's API, copying requirements, and performance expectations before defining tests and what completion means.

Written by the indexing model from the issue text.