ArrayBench broken on intex branch since JIT compiler landed

[mykmelez]

@brendandahl, @marco-c: The ArrayBench benchmark broke on the intex branch when the JIT compiler landed in 4ad785b, even if compilation is disabled:

1.2:1.5 | Time to allocate int arrays of variable lengths a constant 1000 times, by length, in ms:
1.2:1.5 | ArrayBench-1: 12
1.2:1.5 | ArrayBench-2: 1
1.2:1.5 | ArrayBench-4: 0
1.2:1.5 | ArrayBench-8: 0
1.2:1.5 | ArrayBench-16: 1
1.2:1.5 | ArrayBench-32: 1
1.2:1.5 | ArrayBench-64: 1
1.2:1.5 | ArrayBench-128: 1
1.2:1.5 | ArrayBench-256: 1
1.2:1.5 | ArrayBench-512: 2
1.2:1.5 | ArrayBench-1024: 1
1.2:1.5 | ArrayBench-2048: 3
1.2:1.5 | ArrayBench-4096: 5
1.2:1.5 | ArrayBench-8192: 10
1.2:1.5 | Cannot enlarge memory arrays. Either (1) compile with -s TOTAL_MEMORY=X with X higher than the current value 134217728, (2) compile with ALLOW_MEMORY_GROWTH which adjusts the size at runtime but prevents some optimizations, or (3) set Module.TOTAL_MEMORY before the program runs.
1.2:1.5 | Cannot enlarge memory arrays. Either (1) compile with -s TOTAL_MEMORY=X with X higher than the current value 134217728, (2) compile with ALLOW_MEMORY_GROWTH which adjusts the size at runtime but prevents some optimizations, or (3) set Module.TOTAL_MEMORY before the program runs.

[marco-c]

For some reason, $.ctx.nativeThread.nativeFrameCount is always 1 during the test, so we're never collecting garbage.