We can always use multiple pins to get more current into the box if needed. D sub connector will have quite a few spare pins.
There exists
one completely sealed USB socket at Altronics that I know of but it is quite bulky. It would probably fit... just... with the latest inverse PCB orientation.
I can see a possible problem with the new inverted PCB design and the alignment of the sensor. Unless the lid is tightened metal-to-metal the alignment will change every time the camera is opened. I don't have any diecast boxes here to check, do you know if the lids get tightened all the way?
I keep thinking if it would be possible to use the 4 screws which hold the PCB to change the PCB and hence the sensor alignment. Drill the holes through, and then spring-load the PCB somehow... not sure how... but the idea was to turn the screws from the outside to change the alignment of the PCB and hence the sensor.
Looking at Filip's design, I think he will seal the sensor area. The red plate that the peltier sits on has only cutouts for the posts for the feet from the sensor to go through. A bit of silastic around the posts will seal this nicely. I think I saw somewhere that he is using multiple stacked posts to gain height (and reduce heat transfer?). This way he will even be able to leave one row of posts glued in while being able to separate the PCB from them.
Then there is something that looks like an orange seal around the Peltier.
Here is a question. If the whole box is sealed and moisture inside controlled carefully (dry gas and desiccant and possibly a humidity sensor), is there any need to keep the sensor in a separate dry chamber?
This is an interesting post about the sensor temperatures, in particular see Richard's comments and PDF. HQY8 has the same sensor as we are using.