@kha I implemented a more complex example for the paper. The difference
in performance is awesome. On my Linux desktop:
Lean time: 3.34 secs
OCaml time: 11.72 secs
I believe that the difference in performance is due to destructive
updates happening in transformation functions such as `reassoc` and
`const_folding`. I will add a flag to Lean to disable `reuse/reset`
automatic insertion.
BTW, this test requires `ulimit -s unlimited` to avoid stack overflows.
