Given BLE's asymmetric roles and S110's limitation, it's very difficult to do multi-hop on top of it.
If you want to use the nRF51 chip with some other symmetrical, non multi-hop protocol stack, yes, it's possible to do some data forwarding in the application. Don't underestimate the difficulty of a full mesh routing solution; you can waste a whole year on it. Tree routing is better if your application can tolerate higher latency. It's much simpler. The one used in ZigBee 2006 and non-PRO version of ZigBee 2007 is a good reference. Even so, you're likely to spend a few months.
Practical suggestion: stay away from wireless mesh protocol designs unless it's your job. Use ZigBee or other mature protocols with such capability. They often come free with the chip, with documentation and support.
Bluetooth 5.0 might support meshing... who knows?;)
But this is highly unlikely. The initial establishment of a mesh can take upto a few seconds before all wireless nodes have discovered their neighbors, and first data transfers will take additional penalty of initial route discovery. This is very much against the purpose of a mobile, ad-hoc networking technology which Bluetooth is known as.
Nordic may want to expand into new application spaces, i.e. wireless sensor and control network. I think their ultra-low-power radio when combined with energy harvesting will be appreciated in many applications. Right now, Bluetooth is the mass market, and BLE is the largest jewel on the crown. So I won't hold my breath.
