Circular orbit around a point that is not in the center
We can actually find the force law which produces any curve $r(\theta)$ using the Binet equation, as outlined in my previous answer. In this case, we have $$ r = r_0 \cos \theta + \sqrt{a^2 - r_0^2 \sin^2 \theta} $$ where $r_0$ is the distance from the origin to the center of the circle (displaced along the $x$-axis), and $a > r_0$ is the radius of the circle. After a significant amount of algebra (confession: I used Mathematica), we find that we must have $$ F(r) = \frac{8 a^2 L^2 r}{\mu (a^2 - r_0^2 + r^2)^3}, $$ which implies a potential of $$ U(r) = - \frac{2 a^2 L^2}{\mu (a^2 - r_0^2 + r^2)^2}. $$
It may look a bit odd to have the properties of a particular particle's orbit in a potential that's supposed to apply to all particles. A better way to look at this is that such trajectories are possible if the potential is of the form $$ U(r) = - \frac{k}{(b^2 + r^2)^2}, $$ But there are constraints on the shapes of the eccentric-circle trajectories; for a fixed potential, only eccentric circles with $a^2 - r_0^2 = b^2$ are possible. Moreover (as was pointed out in the comments), most trajectories in this potential will not be eccentric circles, and in fact will not even be closed orbits. The eccentric-circle trajectories arise only for special initial conditions; in particular, a particle describing an eccentric-circle trajectory with parameters $a$ and $r_0$ must have $L^2/\mu = k/2 a^2$.
For those who are interested, I have provided Mathematica code to do numerical integration of this potential. It certainly looks like an eccentric circle.
F[r_, bb_, kk_] = -D[-kk/(bb^2 + r^2)^2, r]
a = 2; r0 = 1; b = Sqrt[a^2 - r0^2]; k = 1; l = Sqrt[k/(2 a^2)];
tmax = 100;
soln = NDSolve[{r''[t] == F[r[t], b, k] + l^2/r[t]^3,
ph'[t] == l/r[t]^2, r[0] == a - r0, r'[0] == 0, ph[0] == 0}, {r[t],
ph[t]}, {t, 0, tmax}]
ParametricPlot[{r[t] Cos[ph[t]], r[t] Sin[ph[t]]} /. First[soln], {t,
0, tmax}]
Let $d=2a$ be the diameter of the circle, and let $d=f+c$, where $f,c>0$.
$f$ will be the distance from the center of the central potential to the point on the circle farthest from it, and $c$ the distance from the center of the central potential to the point on the circle closest to it.
Note that the angular momentum $L$, and thus areal velocity (which is $L/2m$), is conserved for any central potential, and that $v_f f= v_c c = L/m$, where $v_f$ is the speed at the far point and $v_c$ is the speed at the close point. (We don't have to worry about cross-products here, because the orbit curve is perpendicular to the vector from the attractive point to the orbit curve at these two points.)
If you do a little bit of thinking, you will see that $v_f$ is the minimum speed and $v_c$ is the maximum speed. Because the orbiting object is always increasing its distance from the attractive point as it moves away from the close point, the attractive potential will slow it down until it reaches the far point, at which point the attractive potential starts to reel it in, increasing the speed again.
Now you can compute the area of the circle from $\pi a^2$, and divide by the areal velocity $L/2m$ to find the orbital period.
(I'm not sure what the kinetic moment means in this case, I'll edit this answer if you can explain its definition to me.)
We can tell this does not follow an inverse distance potential, because it is a bound orbit that does not form an ellipse with the orbit center at one of the focii, and thus is not from classical gravitation or from Coulomb's law. (There are several proofs of such an elliptical orbit on the web.) A circle is an ellipse with both focii at its center, but since the center of the central potential is not at the center of the circular orbit, the orbit fails the test to be due to an inverse distance potential.
This answer addresses the question of "how could this possibly happen?". The answer is that although it may seem physically contrived today, in fact for thousands of years this sort of phenomenon was actually invoked in an essential way in our best theories of astronomy! (And at the same time, even the astronomers using it sometimes worried that it seemed contrived.)
The thing that really needs to be said here is that this setup is called circular motion around an equant, and it was actually used, along with the more famous epicycles, in Western astronomical models from the time of Ptolemy (~200 BCE?) up until Kepler (~1500 CE). Remember that before Newton, astronomical models were (from our perspective today) basically just pure geometry, motivated by "aesthetic" considerations like "circles are perfect, and heavenly bodies are perfect, so naturally heavenly bodies always exhibit uniform circular motion" rather than what we'd recognize as principled physical theory today. (This is obviously an anachronism -- in those days there was really no such thing as "principled physical theory" as we know it today -- so what else were they supposed to do?!) Equants and epicycles were essentially geometric constructions which could be sorta, kinda motivated from these principles, used to overfit the observational data, which was precise enough to make its deviation from simple circular motion apparent (not to mention that geocentrically, epicycles were basically required to explain retrograde motion). My understanding is that for the thousands of years they were used, there was always a certain amount of unease with "aesthetic" acceptability of epicycles and equants. But nobody came up with anything better, so they stuck.
I'm sure I've been told that even Copernicus needed to use epicycles since he didn't have ellipses like Kepler. Maybe the fact that he still needed them was even a sticking point in widespread adoption of his heliocentric ideas? I'm not sure, but I wouldn't be surprised if Copernicus also used equants. And to hammer home the point -- even Copernicus and Kepler were pre-Newton, and their models were still what we'd think of as "pure geometry" -- no real physical mechanism or anything. They just managed to come up with new "aesthetic sensibilities" leading to geometric constructions which fit the data more naturally. In fact, I suspect it would be fair to say that Copernicus and Kepler's motivations for their models stemmed partly from precisely this sort of "aesthetic unease" that epicycles and equants tended to elicit from people.
Anyway, when you work through the geometry, just appreciate that for thousands of years, every apprentice astronomer in the Western world churned through the same geometric considerations that you're thinking about now, and then applied them constantly for the rest of their careers!