Implement get_pixel_coords utility for forward kernel

[horizon-blue]

Part of MET-36

Summary of Changes

This PR implements the get_pixel_coords utility function, which will be used in the forward kernel to compute the subset of pixel coordinates that each thread is responsible for based on the thread & block IDs. As part of this, I'm also adding a int_ceil_div utility (essentially int(ceil(a/b))) to handle the potential edge cases where the number of pixels isn't divisible by the number of threads.

Test Plans

I've added some basic tests to check that get_pixel_coords guarantees that each thread is covered by at least one thread (though it might be possible for multiple threads to write to the same pixel when the job is not divisible). To check them out, run

pixi run ctest

[linear]

MET-36 Integrate and test the forward pass

Definition of done:

• the results of render_fmbs function calls should agree with the JAX code up to numerics
  • generate a teddy bear image
• [ ]