Munchausen timing for Julia is probably off

[mauro3]

I ran the https://github.com/JulesKouatchou/basic_language_comparison/blob/master/Julia/test_munchausen.jl script on my laptop and got times of 7s to 5s for Julia 0.6.4, 1.0 and 1.5. Thus the 100s on a Xenon-node seem off.

[DNF2]

In the Java code, there is an early bail-out when the intermediate sum exceeds the input number. This bailout does not occur in Julia, and not in some other languages either, and makes a big difference.

Furthermore

digit = thisnumber % 10
sumofpowers += power_of_digits[digit+1]
thisnumber ÷= 10

should use divrem instead of first doing rem and later div, which is redundant.