fix: add proof for the std::pmr::vector that it only use the stack space buffer instead of heap

[YukunJ]

Brief: Augment the code sample in std::pmr::vector section.

The tutorial states "vector 的内存来自栈上 buffer，零全局堆分配" without actually prove it.

QA:

see godbolt reference: https://godbolt.org/z/WxjTb8oW5

@Charliechen114514

[Charliechen114514]

@YukunJ

感谢贡献!（最近才知道可以@人提示，手动捂脸）

的确这个地方少了一部分论证。您用地址范围比对把这点坐实了:v.data() 落在 [buffer, buffer+sizeof(buffer)] 内,论证清晰,附的 godbolt 链接也很贴心！

PS: 刚好这边就在小重构Vol3的内容，13-custom-allocators.md 之后会迁移到一个子目录中！届时我会把这段证明一起带到新位置,不会丢。;)

Thank you for your contribution! (I only recently learned that you can @ someone for a prompt)

Indeed, this part of the argument was lacking. You have confirmed it using address range comparison: v.data() falls within [buffer, buffer + sizeof(buffer)], making the argument clear. The attached Godbolt link is also very considerate!

PS: I happen to be doing a minor refactoring of Vol3 here. 13-custom-allocators.md you changed will be moved to a subdirectory! At that time, I will also move this proof to the new location, it won't be lost ;)