Just wanted to add a comment that (in my opinion) the test procedure for icing that was set up to replicate in some way the combination of limiting permissible conditions was the best way to go. It may be true that it went a little beyond the actual conditions (as they are thought to be known). But it is well to consider what can happen.
I read the test reports carefully; they seemed well thought out. The only thing which caught my eye was that it had never occurred to me that the fuel piping and heat exchanger could be such an efficient scavenger of small amounts of water from the fuel.
One thing I saw way back was a Boeing memo concerning the use of the 3-degree Centigrade safety margin in using the stagnation* temperature of the outside air as the limiting low temperature of the wing interior.
*A term from thermodynamics; yes, I know a different term is usually used here. (a senior moment that I cannot recall it).
Anyway, 3C margin is a round number: 5 degrees in Fahrenheit as we mostly used to use. That margin goes back a long way-- at least to before the mid-60's and the days of the first commercial jet-- I researched that at the time. Interestingly, in the Boeing memo the justification for its continued use was that it had been used for a long time without any problems (or words to that effect). The memo was a quite frank, thorough and informed discussion of the issues surrounding fuel temperature, written before this incident. It was an informative memo of the sort that is so seldom seen in engineering any more. I was going to mention it and post a link to it, but when I did so, it had been removed (or moved).
The moral of this is, I think, that there are things in engineering that we do not fully understand (or lack the capability to deal with using practical mathematics), and so need now to increase our margins in this area.
OE