T2CAS, good as it is, is in no way comparable to TFR as found in Tornado IDS, F-111, B-1, etc... anyone that thinks that is, well, nuts. I'm not sure anyone actually said this, but I am getting a vibe that a few think you can ride it in a similar way.
While a B-1 at some 600MPH+ with 'hard ride' selected might make it through such a mountain range with less than 500ft clearance, it is wholly reliant on the fidelity of the terrain database and based on that alone TFR will figure the safe limits. And remember this is an autopilot mode.
While T2CAS certainly borrowed some elements from TFR technologies, it is in many ways the analogue to a stall warning - the point is to avoid the stall in the first place, once you're in it, then it has failed - same with T2CAS - the point is to keep away from the terrain in the first place - not white-knuckle it through. Somehow I doubt this was what was going on.
A-Van raises a good point on the fidelity of the terrain database in this region. This may be the critical 'hole-in-the-cheese' for this accident. While the crew no doubt were highly experienced, if the T2CAS (TAWS...) only has DTED0 for this area, the warnings become significantly more error toleranced. I am not familiar enough with the T2CAS system to know if the current area of database fidelity is factored into algorithm, but given the serverity of the terrain this may indeed be a case of the warning coming to late or quite possibly not at all (since pole spacing for Level 0 is ~ 900m, meaning the terrain height is only known on a 900m grid) - therefore abrupt terrain features may NOT be in the database AT ALL...