nitro-cli-config leaves CPU pool allocated if memory reservation fails

The following code https://github.com/aws/aws-nitro-enclaves-cli/blob/main/bootstrap/nitro-cli-config#L769-L775
gracefully rolls back memory reservation if there's not enough huge pages available, but doesn't roll back the CPU pool.

Implement rollback for CPU pool as well

Open bootstrap/nitro-cli-config around lines 769-775 and trace the failure path used when huge pages are unavailable. Verify how the memory reservation is rolled back, then ensure the CPU pool is also released on that same failure path. Run the relevant nitro-cli-config flow and confirm neither resource remains allocated after a failed reservation.

Written by the indexing model from the issue text.