We should be able to memoize normalization sequences to improve performance on large terms.
We should be able to memoize normalization sequences to improve performance on large terms.