Nope, traceroute keeps going via the Hub.
Tried and tried again to reset the connection and clear NHRP entry and it comes up with the same result.
BTW, what you are referring to should be DMVPNv3.
In case of DMVPNv2 the route next hop already points to the final destination (the other spoke)
This gets an NHRP request triggered since the source spoke does not know how to reach that next hop (CEF entry will show up as incomplete). Once the requesting spoke receives the NHRP resolution reply, it will now have a complete CEF entry and it will be able to forward the packet.
However, as it is happening above, the NHRP/CEF entry are alrey there, except that they are populated with the wrong NBMA (the hub)....
I guess this is the behaviour that NHS should present when it detects that one of the spoke is behind a NAT, but this is not the case in my Lab.