Hi,
This time I processed three odt files/books. The progress are continuous. The main points are these ones:
- One file with no structure failed to display on my Sony PRS-505 because no splitting was initially done, and none automatically performed with the extension.
- One file with 45 text chunks was very nicely reordered.
- The two TOCs were nicely processed
- This time, I did not have images to process
From this, my conclusion is:
1.
The mysterious "reordering problem" seems to be definitely gone. 
2.
Splitting is not performed when it could be.
So, in these cases, the user needs to either correct its original file or correct the epub file with Sigil. For your information, it was a 283k file, so it would seem that the automatic splitting process should occur lower, maybe as low as 150k to be on the safe side.
Meta-data are well processed without use of Sigil if we take care of using your naming convention. (Author - Work)
Personal wish.
As 6 inches readers have only a small real estate screen space, I really would like to see smaller margins. On the AWP thread, Charleski proposed recently a nice and simple way to deal with margins.
So, now looks like the goal will very soon be reached. Congratulations.