Nemrytter, as I previously noted I'm not an expert, but when we did some Ice Crystal testing on a CF6 several years back, I believe they simply used large blocks of ice, with some sort of device that shaved off ice crystal material - the technique was heavily criticized at the time as not being representative, but no one had good suggestions as to how to make it better
Wow, it definitely isn't representative! As you say though, better than everyone else's (non-existent) suggestions.
One detail, ice crystals themselves do not show well on weather radar - to avoid they need to look down for the convective weather below them that's causing the IC threat rather than looking for the crystals directly.
The other problem is that they can often drift quite a long way from the area that does show up on radar. Off the top of my head we've seen them in satellite data up to around 60km away from the area that would produce strong radar returns. Hence Boeing's 50nm advice, I'd imagine.