Verify DPCPU and VNET pointer relocations in riscv purecap kernels

[markjdb]

See PR #1909. The same problem might need to be fixed on riscv as well. Opening this to keep track of the issue until I have a bit of time to spend on it.