simplifier ticks

Just to write this issue down, here are the simplifier ticks from LULESH. On my laptop it takes about 30 seconds for this program to make it through the front end, which is quite long. A more efficient implementation of weakening seems like an obvious target to speed up this step. We should also just check that everything is being collected that should be.

```
32.731:Total ticks: 175294

1689 Inline
834 prj/Var
548 prj/Tuple
288 prj/Const
11 prj/Let
8 Var
1942 RuleFired
1320 CSE
127 x+0
104 commutes *
97 commutes +
54 x*1
53 mapD
42 generateD
30 -y+x
30 aletD/float
20 aletD/bind
15 aletD/eliminate
8 commutes ==*
7 backpermuteD
7 intersect
7 zipWithD
5 transformD
4 commutes min
4 x/1
3 commutes max
2 applyD/identity
2 reshapeD
1 unzipD
3 KnownBranch 3 True
191 BetaReduce
140 dead exp
51 inline exp
171005 Substitution
169949 weakenE
463 shrink exp
211 inline
154 weaken
86 compose
45 sink1
30 sink
28 constant fold
20 replaceE/!
18 replaceE/shape
1 substitute
464 SimplifierDone 464
```