Thinking about the timestamp overflow issue a bit further, it could have been implemented as a soft fork by exploiting Bitcoin's time warp bug. By my calculations this could extend the lifespan of the existing chain by some 240 thousand years.

However, after the activation of BIP113 in 2016, a time warp exploit is no longer a viable option as there's no way to do make timelocked transactions with a far-future timestamp spendable at the appropriate time.

But that part could be addressed with a forced soft fork at worst. So my point still stands ("tis but a scratch!"): Bitcoin will "never" need a hard fork.