It turns out that the rounding effectively isn't performed where many of us expect (i.e. on the fraction of a second right at the end). Instead, the rounding essentially takes place in the initial calculation, which is computed as an integer (and thus presumably the value returned is truncated). The headstart is scaled up to the same level as the speeds when calculating the intercept point, which means that you get back an integer in seconds, and then you can pull out the number of hours and minutes from there.

I hope that this helps save someone the frustration I had. I don't want to say any more, because I'm worried this is already too spoiler-y.

very clean, I like it!

This Kata is meh. I pass 54 out of 58 tests, 3 of them failing are due to rounding discrepancies in the expected results and one of them having for some reason an expected result that differs from the result one would get with the actual input.

I'm having the same issue, I don't understand how that can possibly become 18h and 20mins.

I like it, very elegant