SSEM emulation is incredibly slow on Windows

[vbmacher]

Tested with Java 14

[vbmacher]

[ERROR] Memory listener error
java.lang.ArrayIndexOutOfBoundsException: Index 128 out of bounds for length 128
        at net.emustudio.plugins.memory.ssem.MemoryContextImpl.readWord(MemoryContextImpl.java:46)
        at net.emustudio.plugins.memory.ssem.MemoryContextImpl.readWord(MemoryContextImpl.java:26)
        at net.emustudio.plugins.device.ssem.display.DisplayGui$1.memoryChanged(DisplayGui.java:53)
        at net.emustudio.emulib.plugins.memory.AbstractMemoryContext.lambda$notifyMemoryChanged$0(AbstractMemoryContext.java:89)
        at java.base/java.util.concurrent.CopyOnWriteArrayList.forEach(CopyOnWriteArrayList.java:804)
        at java.base/java.util.concurrent.CopyOnWriteArraySet.forEach(CopyOnWriteArraySet.java:425)
        at net.emustudio.emulib.plugins.memory.AbstractMemoryContext.notifyMemoryChanged(AbstractMemoryContext.java:87)
        at net.emustudio.plugins.memory.ssem.MemoryContextImpl.writeWord(MemoryContextImpl.java:60)
        at net.emustudio.plugins.memory.ssem.MemoryContextImpl.writeWord(MemoryContextImpl.java:26)
        at net.emustudio.plugins.cpu.ssem.EmulatorEngine.writeInt(EmulatorEngine.java:112)
        at net.emustudio.plugins.cpu.ssem.EmulatorEngine.step(EmulatorEngine.java:82)
        at net.emustudio.plugins.cpu.ssem.EmulatorEngine.run(EmulatorEngine.java:127)
        at net.emustudio.plugins.cpu.ssem.CpuImpl.call(CpuImpl.java:147)
        at net.emustudio.plugins.cpu.ssem.CpuImpl.call(CpuImpl.java:44)
        at java.base/java.util.concurrent.FutureTask.run(FutureTask.java:264)
        at java.base/java.util.concurrent.ThreadPoolExecutor.runWorker(ThreadPoolExecutor.java:1130)
        at java.base/java.util.concurrent.ThreadPoolExecutor$Worker.run(ThreadPoolExecutor.java:630)
        at java.base/java.lang.Thread.run(Thread.java:832)

[vbmacher]

Slowness was caused by imprecise LockSupport.parkNanos() on Windows. The exception was caused by wrong handling of memory addresses. The commit above also improved measuring how many nanos we should wait after each instructions to get 700 instructions/second.