Common: Set a min size on threadpool work chunks.

This should avoid slicing loops, etc. into too many chunks.  Generalize
the memcpy a bit.

Core/ThreadPools.cpp

@@ -6,9 +6,14 @@
 std::unique_ptr<ThreadPool> GlobalThreadPool::pool;
 std::once_flag GlobalThreadPool::init_flag;
 
-void GlobalThreadPool::Loop(const std::function<void(int,int)>& loop, int lower, int upper) {
+void GlobalThreadPool::Loop(const std::function<void(int,int)>& loop, int lower, int upper, int minSize) {
 	std::call_once(init_flag, Inititialize);
-	pool->ParallelLoop(loop, lower, upper);
+	pool->ParallelLoop(loop, lower, upper, minSize);
+}
+
+void GlobalThreadPool::Memcpy(void *dest, const void *src, int size) {
+	std::call_once(init_flag, Inititialize);
+	pool->ParallelMemcpy(dest, src, size);
 }
 
 void GlobalThreadPool::Inititialize() {