Replace all uses of AMTL threads with STL threads.

This also rewrites the work loop for threaded queries. It has been
simplified significantly.

extensions/clientprefs/extension.cpp

@@ -297,7 +297,7 @@ void ClientPrefs::DatabaseConnect()
 
 	// Need a new scope because of the goto above.
 	{
-		AutoLock lock(&queryLock);
+		std::lock_guard<std::mutex> lock(queryLock);
 		this->ProcessQueryCache();	
 	}
 	return;
@@ -310,7 +310,7 @@ fatal_fail:
 bool ClientPrefs::AddQueryToQueue(TQueryOp *query)
 {
 	{
-		AutoLock lock(&queryLock);
+		std::lock_guard<std::mutex> lock(queryLock);
 		if (!Database)
 		{
 			cachedQueries.append(query);
@@ -328,8 +328,6 @@ bool ClientPrefs::AddQueryToQueue(TQueryOp *query)
 
 void ClientPrefs::ProcessQueryCache()
 {
-	queryLock.AssertCurrentThreadOwns();
-
 	if (!Database)
 		return;
 
@@ -373,7 +371,8 @@ void ClientPrefs::CatchLateLoadClients()
 
 void ClientPrefs::ClearQueryCache(int serial)
 {
-	AutoLock lock(&queryLock);
+	std::lock_guard<std::mutex> lock(queryLock);
+
 	for (size_t iter = 0; iter < cachedQueries.length(); ++iter)
 	{
 		TQueryOp *op = cachedQueries[iter];

extensions/clientprefs/extension.h

@@ -37,8 +37,8 @@
 #include "smsdk_ext.h"
 #include "am-vector.h"
 
-#include <am-thread-utils.h>
 #include <am-refcounting.h>
+#include <mutex>
 
 char * UTIL_strncpy(char * destination, const char * source, size_t num);
 
@@ -159,7 +159,7 @@ public:
 
 private:
 	ke::Vector<TQueryOp *> cachedQueries;
-	ke::Mutex queryLock;
+	std::mutex queryLock;
 	IdentityToken_t *identity;
 };