Benchmarks results.

I don't know if the speedup is purely random, if I was just lucky, or if the change really makes Python faster...

spectral_norm is a benchmark highly impacted by code placement.

haypo@speed-python$ python3 -m perf compare_to /home/haypo/benchmarks/2017-02-10_00-20-default-e91ec62da088.json call_ref_e91ec62da088.json -G --min-speed=5
Slower (1):
- spectral_norm: 242 ms +- 3 ms -> 282 ms +- 2 ms: 1.16x slower (+16%)

Faster (15):
- xml_etree_process: 193 ms +- 4 ms -> 173 ms +- 3 ms: 1.11x faster (-10%)
- call_method: 12.4 ms +- 0.6 ms -> 11.3 ms +- 0.3 ms: 1.10x faster (-9%)
- hexiom: 18.5 ms +- 0.2 ms -> 17.1 ms +- 0.1 ms: 1.09x faster (-8%)
- sympy_expand: 940 ms +- 12 ms -> 866 ms +- 13 ms: 1.09x faster (-8%)
- regex_effbot: 5.39 ms +- 0.09 ms -> 4.99 ms +- 0.06 ms: 1.08x faster (-7%)
- chaos: 235 ms +- 2 ms -> 219 ms +- 2 ms: 1.07x faster (-7%)
- unpickle_pure_python: 686 us +- 11 us -> 639 us +- 7 us: 1.07x faster (-7%)
- chameleon: 22.6 ms +- 0.4 ms -> 21.0 ms +- 0.3 ms: 1.07x faster (-7%)
- nqueens: 220 ms +- 2 ms -> 205 ms +- 3 ms: 1.07x faster (-7%)
- telco: 15.0 ms +- 0.6 ms -> 14.0 ms +- 0.2 ms: 1.07x faster (-6%)
- sympy_str: 423 ms +- 4 ms -> 399 ms +- 3 ms: 1.06x faster (-6%)
- scimark_monte_carlo: 223 ms +- 9 ms -> 211 ms +- 8 ms: 1.06x faster (-6%)
- xml_etree_generate: 223 ms +- 3 ms -> 210 ms +- 2 ms: 1.06x faster (-6%)
- xml_etree_iterparse: 192 ms +- 3 ms -> 182 ms +- 4 ms: 1.05x faster (-5%)
- sympy_sum: 191 ms +- 7 ms -> 181 ms +- 7 ms: 1.05x faster (-5%)

Benchmark hidden because not significant (48): (...)