1) If internet infrastructure providers are instructed to attack P2P traffic as a condition of their being allowed to operate and under the guise of protection against crime, terrorism, or disruption of monetary systems, then there is a distinct correlation between the transaction rate and the ability of Bitcoin to function under such constraints. At 7 TPS it is not impossible that Bitcoin could continue to function using such methods as steganography and out-of-band networks. Bitcoin enjoys value in part because it has some theoretical potential to survive a dedicated and forceful attack, and I absolutely don't rule out the potential for such a thing as we move into the future (which is heavily postulated to involve a 'cyber 9/11 event.')
I don't see anything within the hard fork which prevents using out-of-band or mesh networks to process Bitcoin transactions .
The higher the transaction rate, the less likely it will be that reasonable function can be achieved under attack. Below a certain threshold traffic can slip through almost any crack in the wall. The absolute values can only be guessed at since there are a lot of unknowns, but the general principle is pretty easy to understand.
More significantly, if systems and expectations evolve assuming high reliably and high capacity network potential, degradation (or more likely, shattering) of this potential is much more likely to result in a total loss of the system.
A system which remains at a defensible data rate allows the opportunity for the needed defensive solutions to occur, though only a relatively few people will be interested in working on them. If the overall system is operating at a data-rate based on full and free use of commodity network capacity then it would be impossible for those who might be interested in network hardening to run the native system.
Total bulk data is also a consideration, but not as significant as network issues. At some point it will be necessary or at least more practical to jumpstart a full node with physical media one way or another.