One thing that I had not mentioned when I was finally convinced the heat entered through a crack/hole in the RCC and heat buildup from the interior in the wheel well.
We all are talking of the shedding of the tiles, I concur that tiles were shed. But look at the cross section of the tiles below.
Wouldn't the heat from the interior of the wing cause excessive shedding?
This is why the foam strike a single critical tile could destroy the spacecraft. The eddy effect described earlier could deepen a smashed-in cavity on the affected tile and either cause a small breach in it with plasma hitting the aluminum wing surface, or just reducing its insulative capability a whole lot. The aluminum skin underneath heats up, the adhesive for the affected tile loosens and then the whole tile comes off, and suddenly there is a bigger aluminum area exposed to the plasma and the unzippering effect starts.
An RCC breach was not needed to create the known evidence, other than the USAF photograph. The extra heat from Columbia's re-entry profile for this mission, its overweight and transition turbulence causing excessive drag (the latter happened on a dozen prior missions!) alone could have caused its loss. But there were detachment events starting no later than the Owens Valley indicating at least a tile breach much earlier in re-entry so I always thought, with one brief wavering when Xbob mentioned the elvon/elevon burn-through he personally inspected, that the foam strike had fatally damaged some part of the re-entry protection.
Until just recently I thought it was the TPS, specifically the tile between the RCC and the port wing landing gear cover. John Jamieson and Xbob showed me how tile loss could be explained by an RCC breach which was not immediately fatal. The USAF photo and their interpretation of it (I discount speculation by the clueless media but definitely consider some interesting guarded comments by NASA) got me to buy their theory of an initial RCC breach.