Same thing, I have 8s runtime, but none complex tests passed. It seems this issue will never be fixed as one can find this kind of reports in a lot of katas in relation to C++ solutions.
In this case it seems even the type in Equals that is created by the test itself fails.

Your solution is compiling and running on Codewars too. It's just that it's getting the answer wrong -- and the Codewars test framework is reporting this rather ungracefully, because it doesn't seem to know about tuple types in C++.

It would be better if there were specified limits for m and n, in C++ arguments are long long and that's quite a large limit. With my current solution I get around 8s for [1, 1000000] range and this passes the tests, but it would be more useful to have input range and time limits specified as a hint from beginning. In other systems similar to codewars when I see specific range and time limit I get an idea of what direction I should take (in some systems there is even specified desired complexity), thus if a simpler solution will suffice or I need to come up with a more complex one without wasting time on trying this out with tests.