[popo]

Let's say the average RAW file from a 7D (18MP) is 25MB. Scale that, it is about 350MB for 250MP. Higher end SD or CF cards can write over 100MB/s, so we're looking at a few seconds to write each one. It would likely make sense to do parallel writes to multiple cards at this point. The sensor readout works out at 5fps so it will be card write limited. Capacity is not a problem since you could get close to a hundred shots per 32GB card and higher capacities are available.

Usage: for me, oversampling. Look up sample theory if you don't know what that is and why it might be useful. It does require a different mindset than traditionally taken. In short, a huge number of data points (even if noisier) when correctly reduced to desired output will generally look better than just sampling at the desired output resolution. There's a lot of small print I could add to say where it may be better or worse, but it is beyond the scope of a post here.

The pixel density is nothing special. Compact cameras have exceeded this level for years, and latest ones can still give half respectable pixel level output.