So there are multiple things to factor in here:
The 8 hardware lights is just a limitation of old-style directX 8 shaders. Since GM uses DirectX 9 now, we have the freedom to pass in as many lights as we want into shaders. That being said, if we are using forward rendering, this is inefficient as every pixel on screen needs to perform lighting calculations for every light. There are two main ways in which we can optimise lighting:
1) Forward pass lighting: We still pass lights into a shader, but we change which lights are active. This also applies to GM lights. Just because we are limited to 8 lights, this does not mean we can only use 8 lights in our scene, it simply means that we can only use 8 lights SIMULTANEOUSLY when rendering one object. Each time an object is rendered, you can simply place those 8 light points on the nearest 8 lights to that one object. Realistically, you dont even need 8 lights, you only need 3 or 4.
The main downside of forward rendering is that we cannot apply drawing optimisations. If we want to render a whole scene as one model, then yeah, this technique goes out the window. Though it is important to be aware of it.
2) Deferred Rendering: This is the idea that we separate our rendering out into a number of processes. We have to keep a G-buffer, this is a collection of surfaces which contain information about our scene. We will normally use a number of common buffers for light: Diffuse (This is the rendered un-lit scene with textures), Depth (This stores per-pixel information of how far away a given point is from the camera), Normals (This stores per-pixel information about the direction in which the surface of a given pixel is facing.)
We then use a second pass to render lights onto a light buffer. Now we can cheat here, we draw spheres in the areas the light is. This means that lights only do processing for the pixels they affect, and far away/out of sight lights do not have any performance impact on the rendering process (This is not true for forward rendering). The downside is you have to do a little more work to set the system up, though you get other benefits such as the ease of adding in normal mapping or other renering effects.
Deferred rendering can theoretically support 1000s of lights in a scene, and on screen at any one time, the number of lights you can render before you hit a fillrate crash, though on a modern GPU, that'll be in the 100s atleast.
The final step is to merge the light surface with the diffuse.
Now, I understand this is all quite hard to grasp at first, though learning about 3D rendering techniques is a fantastic and fun skill to have. The forward lighting approach is what is built in and being smart about which lights we draw at any given time can be really helpful. Unfortunately, I can't really give you any help in setting up a deferred renderer, as its something that requires your own understanding in order to be able to use well. GM currently has no system/pipeline which abstracts away all the difficulty of it. I was working on one, but stopped to focus on my main project.
I can walk you through the process, though it'll require that you have some knowledge of shaders + surfaces in order to achieve it successfully and to really have control over what you are trying to do.