good Kata!

There are no units mentioned at all, or only at some points.

I am not sure what units exactly you mean, but some non-SI units used in the kata are customary to the domain of the task: (i.e. cycling): times tend to be expressed in minutes (pace is min/km) or in hours (speed in km/h or even mph), hill gradients are percentage, etc.
Some of the non-SI units used in the kata are IMO ok, but still every value should be described in full, including associated units. For example, I am totally not sure what the GRAVITY_ACC = 9.81 * 3.6 * 60.0 means.

I always think the same, haha. xD

C translation has random tests exit at the first incorrect answer. This deprives the poor solver from much needed data points to guess their way through this kata.
Please adapt to run all tests with outcomes for each test, and report inputs and expected outputs in case of errors as well.

As others have already noted, the description is severely lacking in several aspects:

1. units are missing
2. unclear physics, author assumes too much pre-existing knowledge

I find the description too long. The whole part of suggested variable names can just be deleted.

And if this is not affecting the challenge too much (I don't think so given the number of actual solves); change the units to SI units!

C translation does not report actual and expected values in tests, only The expression (act) == (expr) is false.

It was easier to earn points in the past :D

It's an old kata.

How is this a 5 Kyu kata? More like a 7 Kyu one it seems.

If you know the algorithm, this is trivial, if you don't, it's very hard to come up with a proof why the simple algorithm works.

The description mentions, but does not show an example (at least for C).

C translation has sample test:

do_test("(((", -1);

How can you have a negative number of reversals? That needs explanation in the description.

You can do it inplace in two passes, or in one pass if using a buffer like I did.