Generalize `Bittide.Hitl.FpgaIndex`

[kleinreact]

Currently, bittide-experiments defines https://github.com/bittide/bittide-hardware/blob/703b881e8e468e7916dc01b2f8c49ba559ac4649/bittide-experiments/src/Bittide/Hitl.hs#L94-L96 being fixed to work for 8 node topologies, but the number of FPGAs available may change and should be configured in bittide-instances instead: https://github.com/bittide/bittide-hardware/blob/703b881e8e468e7916dc01b2f8c49ba559ac4649/bittide-instances/src/Bittide/Instances/Hitl/Setup.hs#L21-L22

Therefore, it may be desirable to generalize the definition in bittide-experiments towards an arbitrary number of FPGAs or to implement the proposal given in the comment.

Note that bittide-experiments cannot just use the value from bittide-instances, as this would introduce circular dependencies.

[martijnbastiaan]

This might be a little bit much to fall into the scope of this particular issue, but I wonder how easy it would be to read a specific test setup from configuration files. E.g., CI would have a setup with 8 nodes, while smaller setups would only have 3. Of course, not all topologies would be testable on a 3-setup, but the code generating test cases could account for that. Most other tests (tranceiver up, riscv test, etc.) shouldn't have to care all that much.

[kleinreact]

Check out the changes introduced by mini-bittide, where you can see what is necessary for changing the current setup to work with 1 FPGA/2 Nodes. It shouldn't be that much effort to generalize that to work towards n FPGAs/m Nodes setups as well.