The open-coded task_vma iter added earlier in this series allows for
natural iteration over a task's vmas using existing open-coded iter
infrastructure, specifically bpf_for_each.

This patch adds a test demonstrating this pattern and validating
correctness. The vma->vm_start and vma->vm_end addresses of the first
1000 vmas are recorded and compared to /proc/PID/maps output. As
expected, both see the same vmas and addresses - with the exception of
the [vsyscall] vma - which is explained in a comment in the prog_tests
program.
Signed-off-by: Dave Marchevsky <davemarchevsky@fb.com>
Signed-off-by: Andrii Nakryiko <andrii@kernel.org>
Link: https://lore.kernel.org/bpf/20231013204426.1074286-5-davemarchevsky@fb.com