1. Do you feel that a group of high school students could complete a polished game in 16 weeks, meeting everyday for about 70 minutes?
Recreating a classic arcade game should be easily doable in this time frame, although I don't think the polish is entirely necessary. Getting the gist of a game is probably the more important thing, if they remake Asteroids with the ship as a square and the asteroids as circles I would consider that a success.
2. The students all want to do 3D games, my principal wants to do 3D, however I feel that producing a "good" game will generally be much harder if we go with 3D. Especially, because we don't teach 3D modeling at my school so I'd have to teach that as well, however most of my students will have had graphic arts and will have Photoshop experience. Do you agree with that sentiment?
I don't think the students will want to do 3D once they see how much work it can be, however, if they think they can tackle it you can always encourage them to do the research for it outside of class. It all comes down to how much effort they want to put in their project. I would recommend that you focus on 2D though as that would be easier to grasp and will allow you to cover aspects specific to programming.
I recommend taking a look at
http://fivedots.coe.psu.ac.th/~ad/jg/ for more information on java 2d game programming.
On a side note, my high school offered a java programming course, it was only a trimester and most people did not know any programming but using Greenfoot most were able to create a mostly functional game in about 3 weeks for our final project. I don't really know how much time people worked on it outside of class though.