python new test framework is required. updated in this fork

Hi,

opening as issue to avoid approval before a decision is made:

I have (and some other users as well / Note: not the published version of mine, it's the one in the fork) a solution that works consistently with 5 guesses at most. I'd suggest to reduce the limit to that number, if that can forbid the somewhat "cheezy" approaches, like the second python solution. Thing is, I'm not sure this will totally forbid the approach, but that might make it tedious enough so that users will go to the expected solution.

If so, that would make a good 4 kyu, once some additional tests/constraints are added (see discussions below)

In any case, the tests must be rewritten (in python at least):

• add the fixed tests first
• the random tests must be wrapped in an it block

Cheers

Keep the number of tests (similar), but make sure to test some of the numbers multiple times, to avoid solutions like mine...

Maybe add a few more numbers to the sample tests, from the top and bottom of the possible range.