It's not just the 3d modelling that needs to be done to add motorbikes though. Each one needs programming time for physics, taking programmers away from being able to do other parts of the game. No one said it CAN'T be done, just that it shouldn't take precedence over the parts of the game that are severely lacking currently.
They could already add endurance racing and standing starts, any programming that would have to be done for those was done in gt5. They removed those things from gt6, but the underlying code would be there, so I don't think those are an issue. The abolition of standing starts from most events was due to the AI being so bad that they needed to stack the deck so heavily in the AI's favour in order to add any kind of challenge to the player, that's why AI should be re-done from the ground up, to bring it into line with current sims.
A flag system would require re-doing the penalty system, which would be good, as the penalty system in gt6 is one of the worst I've experienced. Not sure how much work that would take.
The physics of the cars isn't much better than gt5, and it's riddled with bugs, so in a lot of ways it's worse, so I would expect them to want to re-do most if not all of that for the new system. This is the part that would have their programmers time tied up the longest I would imagine, as it'd have to be pretty damn complex.
Another point would be, the majority of people aren't a fan of having the ps2 models in a ps4 game, so that is a metric **** ton of work if they are wanting to model as much of those as possible in a "premium" quality. If I could choose between bikes in gt7, with not many new cars, and all the standards from gt6, or a higher new car count, and a large portion of standards converted to premium, but no bikes, I would choose the latter. The odd looking ps2 models on the nicer looking ps3 tracks bothers me a little, I can only imagine how out of place they'll look in a ps4 game.
Sorry about the big post.
I get that you're late to the party and all, but let's just recount the facts.
PD made TT by modifying GT4.
Kaz said he's really pleased that people want another TT game, before GT5 launched.
Kaz teased "I can't say no" regarding bikes several times, lastly within a year or so ago.
So, if bikes are coming, they're already a long time coming.
Now, if they have planned so far ahead, and based on their experience with the first retrofitting of a GT to make a TT, they'll be trying very hard to consolidate the development process to minimise, not necessarily the cost aspect, but certainly the total time aspect. Their slow but determined approach to content production allows plenty of time for the volatile, imperfect task of programming - which they only have to do once with the consolidated approach.
As you say, the major component (that isn't content) is the physics.
Physics, though, is physics: it applies to bikes just the same as it does to cars.
So any physics engine that works for bikes can be made to work for cars, and vice versa.
So if you're going to revamp the physics model, why not do it in a general way that benefits both vehicles (if you have plans to include both vehicles at some point)?
Now the cost potentially goes up (because the detail in the model is higher), but so does the benefit to the car simulation - besides, PD's budget must be growing, because its staff is; those decisions will already long have been made. I encourage you to work out for yourself how the necessary features for bike simulation will improve the car simulation, but it's explained in this thread twice at least.
So, since you say that bikes do not affect AI, flags, standing starts and endurance racing, all that's left is sound.
PD have spent several years working on a general system that can account for the wide array of differences present in the diverse car roster in the game; that naturally extends to bikes, which would require no special treatment. But can you also see the parallels in developing a general sound approach to cover content diversity and the need for a general physics approach to cover the content diversity? The other common factor is the timescale.
What else is there?