If your stack pointer is aligned to a cache line when a function is entered, execution of the functions is likely to place less stress on the cache.
So, a compiler system can be organized to insist that functions are entered with the SP aligned on a cache line, and the compiler, knowing how much stack has been used at each call site, how much it takes to re-align the SP before it makes the call.
That would explain your example. I haven't seen many compilers actually do this, since the stack itself tends to grow/shrink by modest distances and overall doesn't put much demand on the cache by virtue of reusing that local storage again and again.
Another use is allocating space for the callee to work in, or to return a result bigger than what fits in a register. A compiler wouldn't write a comment like that, so some person likely did this. Maybe he knew what he was doing; maybe he didn't. If the called function doesn't need this space, then its just a wasted instruction.