neutral stability means the CG is at it's aft limit area to save fuel but it needs a computer to keep it stable
Not quite right. For a conventional aircraft (and I presume this is what the Airbus becomes in the event that ALL computers roll over and die), the aft CG limit may (amongst other requirements) be limited to give an acceptable positive static stability .. neutral stability would be a fail and require either a further constraint of the aft limit or another flight test engineering fix.
It follows that, if the CG is sufficiently far aft to result in neutral stability, then it wouldn't be certificatable or, perhaps, the probability of box failure demonstrated at certification is sufficiently low to be acceptable.
However, this latter consideration may be tied in with the present mishap - inextricably. Are we not all concerned with the possibility that this crash may have been involved with events so out of left field (but only because of presumptions regarding training and competence standards) that the original certification program may not have considered them at all ?
I don't know what the certification story was for the Airbus FBW machines but would sure be interested to find out ..
A pilot would have a hard time to keep it at a constant altitude
True, but that's only a part of the problem. The typical pilot, faced with a sudden deterioration in the stability he/she feels/sees is more likely to rip the wings/tail off while trying to figure out what might be going on ..
a good pilot could still fly it by hand but it would be difficult
More a case of needing to recognise the problem and then be aware of/trained in the changed techniques required to fly the aeroplane. TPs cover this during their training course and there are documented cases of the occasional TP having to drag a statically unstable aircraft around the patch to get it back onto the ground .. but for Captain Average Joe Pilot .. the story, in all probability, is going to have an unhappy ending.