Benchmark: timing 10000 iterations of for1000, for10000, foreach1000, foreach10000...
   for1000: 59 wallclock secs (53.75 usr +  0.03 sys = 53.78 CPU) @ 185.95/s (n=10000)
  for10000: 676 wallclock secs (537.40 usr +  0.23 sys = 537.63 CPU) @ 18.60/s (n=10000)
foreach1000: 33 wallclock secs (24.67 usr +  0.00 sys = 24.67 CPU) @ 405.27/s (n=10000)
foreach10000: 293 wallclock secs (250.74 usr +  0.21 sys = 250.95 CPU) @ 39.85/s (n=10000)