Stack cleaning between trusted and untrusted code

[globusharris]

R exhibits the following behavior:

foo <- function(){
  x <- 5
}
y <- foo() #y is assigned to 5

This could mean that things that we don't want to be on the stack stay on the stack. Should make sure this isn't happening over divide between trusted and untrusted portions of the codebase. Probably impossible to fully guarantee but could make sure that at least a confused researcher doesn't accidentally publish things that they shouldn't.