The inductance is measured in Henrys (H). So, the inductance is 90mH with the relay closed and 60mH with the relay open. As the relay closes, the magnetic path changes, and the inductance changes.
You could use this to predict the behavior of the relay coil voltage when the relay is energized or de-energized. You could use it to make sure the energy capability of a clamp is sufficient. You could use it to predict how smooth a current would result from a PWM energization at a certain frequency or a full-wave rectified energization at a certain frequency.
For example, consider the TPIC6A596. It has a clamp rated to withstand a single-pulse avalanche energy of 75mJ. The worst-case energy stored in the magnetic field of that relay is \$(0.067A)^2 \cdot 0.09H\$ = 0.4mJ.
The change in the inductance as the relay opens means that that energy will be dissipated in the relay coil (mostly) if you use a diode across the coil. That means that the relay contact opening is slowed just when you would want it to be fast in order to minimize arcing. From a page here is a 'scope capture of this effect.
The step plot is the relay contact opening. The bumpy plot is the current through the armature, as you can see it is not monotonic and actually increases as the clapper leaves the armature, which is just as the contacts are opening.
With the numbers from the datasheet (in addition to the typical drop-out voltage and open/close time) you can begin to model this effect, and predict the results of using a Zener clamp vs. a resistive clamp vs. an RC snubber vs. a diode clamp.