Interesting question.
Is the BS 476 / BS EN 1364 test "realistic" ? no it isn't and it was never meant to be so. How could you ever develop a test that would replicate any and all possible fire scenarios? It would be completely impracticable. It is, however, reasonably representative of a real fire & the evidence is that it adequately explores the key performance requirements of fire-resisting construction. The whole world tends to use the same time/temperature curve (ISO 834 - though there are others for specialist applications), so it can't be that bad!
Our regulators know this, so they give us reasonably conservative requirements/recommendations (30 mins, 60 mins, 2 hours or whatever - the rating being based upon risk) knowing full well that the fire-resisting construction isn't likely to last for those times in a real fire; it might do more or it might do less, depending on the particular fire scenario. Similarly, with a neutral pressure plane at about 1m height in the test you at least expose a typical 'person-height' element to both positive and negative pressure in the furnace - whereas in a fire it is unlikely to see exactly the same pressure conditions - it will vary principally depending upon fire size, ventilation and compartment height.
Coming back to the issue of letter plates in fire doors - the key issue for me is about the 'good practice' test. If such an installation was reasonably modern & it would significantly compromise the ability of the door to pass a BS 476 / BS EN 1634 test then personally I'd be looking at remedial action. Plenty of companies make fire-resisting letter plate assemblies that'll demonstrably give up to 60 mins F/R in a typical solid-core fire door, so it's probably pretty difficult to claim it isn't reasonably practicable to correct? After all, if the fire performance of the doorsets came into question, then it's not beyond the bounds of possibility for one of them to be removed and tested (Trading Standards have done this a number of times in the past).