[client] overlay/msg: fix race condition in render

[spencercw]

If an overlay is closed with overlayMsg_close, the message can be freed while it is still being used by msg_render, resulting in a segfault. Lock the message list for the duration of msg_render to fix this.

A semi-reliable way to reproduce this is to put a 20ms sleep after the call to ll_peek_head in msg_render. Use remote desktop to stop the Looking Glass host service, then start the client. The 'host not running' message should appear. Now start the host service. The client will now probably crash somewhere inside msg_render because the message it is rendering has been freed by overlayMsg_close.

Not sure this is the cleanest approach, but I've taken after the precedence of ll_forEachNL by adding some more nl methods. Perhaps it would be better to add a new lock in msg.c.

[gnif]

Honestly I am not surprised there is an issue here, the code that handles the messages has not been looked at in quite some time. Thanks for this!