Implement spend_cpu_cycles in assembly

The compiler was unrolling it previously, so we're doing assembly now for consistency.

benchmarks/bench_cpu_work.cpp

@@ -5,7 +5,7 @@
 #include "../src/cpu_work.hpp"
 
 int main(int argc, char *argv[]) {
-  int iterations = DEFAULT_HEALTH_CHECK_ITERATIONS; // Default: 7000
+  int iterations = DEFAULT_HEALTH_CHECK_ITERATIONS;
 
   if (argc > 1) {
     try {