improve handling of competition reward splits that don't sum to 100

[dkent600]

Currently we require that a reward split in a competition proposal must add to exactly 100. Perfect precision. It is hard then to have a close split between three winners, has to be 33-33-34. We would like to allow 33-33-33 with leftovers that will flow back to the dao, and even better, higher precision on the proportions to minimize the amount leftover, allowing us to get closer to a sum of 100.

[jellegerbrandy]

The logic could be this:

When rewardsplit is [n0, n1, n2, ... nx] the individual rewards are n_i / (n0 + n1 + .. +nx)

For example:

rewardsplit [90,10] for first winner 90%, second place 10%

rewardsplit [1,1,1] for getting three equal parts

rweardplit [1,1] is equal to [50,50] -> two equal parts

etc.