Is the output beam profile of the laser diode array, an elliptical spot (like a single laser diode emitter) or does it have elliptical profiles equal to the number of laser diodes that make it up?
You will get a single ellipse equal to the far field pattern of a single emitter.
Think of each emitter as independent. Each produces elliptical angular profile. In the far field (say if you shine it at a wall that is far away compared to the emitter size) each one produces an ellipse. The emitters are all pointing the same direction (very nearly anyway). The beams are parallel. The ellipses are slightly offset equal to the spacing of the emitters. As the ellipses grow propagating to the far field, the spacing of the emitters becomes more and more negligible compared to the size of the growing ellipse. In the far field you get one ellipse. Because the overall size of the array is small (millimeters) and the angles are large (tens of degrees) you do not have to propagate very far before the single emitter ellipse dominates.
It is important to note that fiber coupling tends to increase the beam quality factor. The core size and the critical angle for total internal reflection give a fiber an upper limit for beam quality factor. This can be characterized as the mode size (effectively the core diameter for simple step-index multi-mode fibers) times 2*NA of the fiber (o.k. 2 x arcsin(NA) to be more exact, but if you want to be exact, check how the fiber NA is defined. It usually ISN"T the 1/e^2 point). You cannot efficiently inject a beam which has a worse beam quality factor than the fiber can handle. In fact it often isn't practical to even closely match the fiber mode to the incoming beam. Usually the selected fiber has a significantly larger core x angle capability than the beam that is being injected. For a relatively short length of fiber, what comes out of the fiber might not be much higher beam quality factor than what was injected. However, after a few meters of fiber, particularly if coiled, the beam will couple into all available modes in the fiber and what comes out of the fiber will have the full mode size and NA of the fiber.
Depending on your application, this might be a bad thing, or it might be a very good thing. The fiber increases the beam quality factor and so for optics with a given output aperture size, the divergence will be bigger than if you had used the original source without putting it through a fiber. If your primary concern is maximum radiance in the far field, that is a bad thing. On the other hand, if you are making an illuminator, often you want to flood a fairly large field of view and tight collimation isn't your biggest concern. If the desired divergence doesn't require too large of an aperture for your optics, then the higher beam quality factor is fine. In fact, it might be great. In spoiling the beam quality product, the fiber also homogenizes the beam. At the output of a multi-mode fiber, the fiber core will be nicely uniformly filled. The output lens conjugates the fiber output, so what you get in the far field is essentially an image of the fiber core: a nice round uniform top-hat illumination. If your primary concern is uniform illumination, then the fiber coupling can be a great thing.