Generating crystals on the ground. One engine test centre likened this to having to generate a CB in an engine test cell. High levels of ice/water content (4mm/m^3), subfreezing conditions (requires liquid nitrogen), low air pressure, and the correct mass/airflow through the engine running at cruise conditions with ice protection on.
Research / test flying several years ago, only managed one engine malfunction in 20 plus flights over a two year period. This work identified aspects of the high levels of ice water content, the duration of exposure (cross anvil vs long duration in the cloud mass) and a myriad of varying engine conditions. This was further complicated by large areas of CB uplift and the difficulty in detecting the conditions with WXR. Also, these conditions could be found in temperate latitudes.
Thus there may have been many encounters, but only a few resulted in observable problems.
50nm is a reasonable margin assuming that the errant CB can be identified; and not to forget that the hazard is a multiple engine event not just the failure of one.
Why now: modern engines are built to much finer tolerances – supercritical designs (cf the stall problems due to squashed flies on an aircraft supercritical wing section). Also the fine running margin to be maintained by FADEC; older designs would cough and spit the ice out, new designs are relatively more fragile.
Also, consider that as WXR capabilities advance, do we fly a little bit closer to CBs because we can see them; but where is the ice, the anvil, etc.