Poorly optimized code

[justinethier]

Cyclone is much slower than chicken 5 on the following benchmarks:

• array1
• chudnovsky (bignums)
• compiler
• conform
• destruc
• gcbench
• graphs
• lattice
• pi (bignums)
• quicksort (suspect poorly optimized mutable operations)
• simplex (was there a beta expansion issue here, too?)

I suspect a lot of this is due to how cyclone poorly optimizes code using mutable operations. If that is the case it may be necessary to rewrite portions of the optimizer or take a deep look to figure out what can be done. For now the first step is to verify and update above accordingly.

This is related to issue #344

[justinethier]

In particular, performance on array1 and destruc is abysmal. I think a large part of the problem is our optimizer is extremely conservative when it comes to mutations (EG: vector-set!) and prevents most of the opportunities to optimize such code.