It makes no sense

[Mark Hygate]

Almost all of the forts on Hadrian's Wall are not only attached to the wall but in fact overlap the wall with their three northern gates beyond it so that troops can come out of any of the three to defend against any attack from the north.

Then there were the milecastles every Roman mile apart with two turrets in between each of these at 500 yard intervals and of course the ditch and vallum mounds behind the wall that serve as a defence behind the wall.

Then as mentioned earlier how would the wall have terminated at this point with the wall not attached to the fort, would it have simply been an open ended wall that any enemy could have very easily just walked around for this fort is some five miles or so from the east coast.

Indeed, I quite understand that and assumed that you did too, which is why you had posed the question.

Much further inland having the forts astride the wall, especially if the 'as designed' garrison was in place, makes complete sense - and it all depends on the topography and defence/offence options available. In this case, however, it is completely likely that the Pons Aelius fort has no less than 3 primary defence possibilities. Firstly to defend that end of the wall, cover the bridge and to defend against a riverine assault (5 miles is not that far). To cover all those requirements, having a completely enclosed/defended fort makes more sense and doesn't introduce a vulnerability that could be exploited before relief forces appear.