New late MI images are released as 22 Dec 2004. These images are taken in Sol 124,125 and 162.
One interesting aspect of these images is they show interesting berries, typically with grooves, with the best focus taken on their sequences. Note that only berries are best focused but not the background.
1M139192463EFF2821P2997M2M1.JPG
1M139192209EFF2821P2997M2M1.JPG,
1M139192463EFF2821P2997M2M1.JPG,
1M139194202EFF2821P2997M2M1.JPG,
1M139194614EFF2821P2997M2M1.JPG
These five images are covering the same grooved berry couple, but belong to different MI sequences. On each sequence the only the focused on berries are late. It is interesting that image 2 and 5 on above list show these berries in a small fraction at the edge of frames and this sufficed to included in this batch.
Other images show also grooved berries in similar fashion.
I think this is more than enough to fall in conclusion to they are held for their content a very precise embargo, possibly requested by their own scientific teams.
This also shows their deep curiosities on berries, which never surfaced in PR's

Another "bearded" berry (1M139284552EFF2829P2997M2M1.JPG) is covered in this batch which is not the best focused one in the MI sequence, but including the late one, the sequence still have 4 images, where we can expect 5 as usual.
I think this behavior could be considered logical although disgusting when the real raw image embargo is considered. If the JPG image is not sufficiently detoriated and still shows worthy details deserve to not be published.
We may name this MI batch as "bad luck" as often we say: Damn it, all images of this great feature are blurred! Now we see it is not pure bad luck.