* Restructure gpu compilation execution pipeline * Add compilation server/client infrastructure * Add wgpu executor