Fixed division by zero

[timkoers]

When the --single flag was used, an division by zero was generated in the munit_rand_state_at_most function.
Fixed and tested

[jibsen]

I just had a quick glance at the code, and am wondering if it might be better to check if possible is 1 at https://github.com/nemequ/munit/blob/master/munit.c#L1617 instead?

My thinking is that calling munit_rand_at_most with a max of 0 is a bug, and it's better to fix that bug than to slow down the PRNG for all other uses by adding the check.

[jibsen]

Hmm interesting, I had a recollection of this being fixed before, and looks like the empty range in #30 was it, but firstly that is in the public API, not the internal function used here, and also the fix seems to have disappeared again?

[jibsen]

My thinking is that calling munit_rand_at_most with a max of 0 is a bug

Looks like my thinking was wrong, the ranges appear to be inclusive, so calling it with max equal to 0 should return 0.