I've made a patch for the latest Christmas release. Do not expect anything "big", it's just a few tweaks and bug fixes of mostly some graphical glitches.
Apart from some overlay synchronization refinements and other graphical issues, I have moved restore_kelvin_wb() calls to another task to make WB be restored before movie mode is turned on, fixed movie mode remapping (movie mode was not remapped to the chosen shooting mode), modified the movie log file format to make it easily readable (+ removed the unused "frame" column since I was not able to get the number of current frame and added column for equivalent focal length).
The thing, which I think Alex will hate me for, is that I have commented out or lowered values of some msleep() calls which only slowed the camera response down and were not needed to be present in the code or to be that high.
I was also hoping to fix the bug of camera locking up if you pull the memory card out too early after opening the card slot door but instead I have speeded the ML boot up process up by ~1.3s. I'm starting to think that the card removing bug is a low-level issue. It is not related to the ml_shutdown() function called when card slot is opened nor to saving the configuration file nor to any infinite loop in boot-hack.c, reboot.c, reboot-all.c, config.c, debug.c or state-object.c. And it is even present when ML is turned off in the boot up process, which means that it is created before one of the very first C tasks is running in the camera (my_init_task(), boot-hack.c).
And finally, the patch includes my previous patches of zebras and other overlays so is uses the whole area possible.