Hi, first post, so be gentle...
Having some experience with Industrial Cameras, and having also been down the road of developing our own camera for a special application there is nothing straightforward about anticipating the roadmap of the sensor developers.
The sensor itself is only part of the question, there are also all the supporting electronics providing the clock and reading out the taps from the sensor.
The FPGA (a programmable chip, the brain of the camera) is usually the limiting factor in a sensors actual performance.
If you want to make one picture every five minutes, you can use an older FPGA with just about any sensor you like, so long as it can be reprogrammed to deal with the data.
Increasing the size of the sensor, the data of course increases as well, and the FPGA will eventually hit a wall in terms of bandwidth.
Most sensors out there are already putting their processors against the wall, and it is always a compromise between what has to be done (de-bayering for example) and the output rate required.
In the end, we chose to drop our "in house" camera design and partner with a couple of suppliers.
In Industrial cameras, standards are becoming the name of the game, once we implement the standard, from a software point of view changing out the sensor tech becomes easier.
But all of this would increase the bulk of the design, if you have to have physical interfaces to make certain components upgradable you need to include space for connectors and accessibility to do it... you would end up with a much larger camera, that certainly wouldn't look like an M.
Hope this was somehow informative,
A