Improve fuzzy analysis speed

Some time ago it was implemented that with every update request t5memory saves the TM to the disk, except a parameter is send with the update request, that tells it, not to do so.
So for internal fuzzy analysis we are sending an update request after each fuzzy request. This means, that for every segment that we analyse the TM is saved to the disk. This of course slows down the analysis dramatically. And we do not need to save the internal fuzzy TM at all to the disk.

What needs to be done send flag "save2disk": "0", when sending update requests to t5memory during an internal fuzzy analysis.