This draft is WIP for current experiment of using plonky2 over gpu.
Plan is to port the prover from sideprotocol's plonky2-gpu crate into our starky prover.
https://github.com/sideprotocol/plonky2-gpu/blob/main/plonky2/src/plonk/prover.rs#L239

Current progress:

• Successfully able to test FFT + merkle tree creation over gpu. You can find the test here https://github.com/0xmozak/mozak-vm/blob/vivek/gpu/circuits/src/stark/prover.rs#L546
  To run, execute, in the project root

cargo test --package mozak-circuits --lib --features="cuda" -- stark::prover::tests::test_cuda_poly_batch --exact --nocapture