When you have systems in a grid, Oracle does make it easier to deploy the Oracle software and patch those instances in the grid, rather than perform these actions individually on each grid node. And with RAC, you can add and remove nodes easier than you could prior to 10g. I have not seen anything new in 10R2 that signficantly improves what was available in R1.
With the grid concept, you should have a means of "delivering information to users whenever they need it, regardless of where it resides on the grid" (quote from the Oracle docs). While 10g lets you use RAC to add and remove nodes from a one cluster to another to handle resource demands, it still does not make Information Provisioning a seamless operation. If a user is accessing a RAC cluster and wants the data in another cluster, then Oracle's mechanisms for making that data available are Transportable Tablespaces and Oracle Streams.
It would be nice if the cluster could access another cluster seamlessly, kind of like a cluster of clusters. This is where the true power of grid computing will show the biggest benefits. But Oracle is not there yet in making all of this seamless. You still have to transport data (with TTS or Streams) and then worry about handling changes to that data. Oracle is not alone in their position on grid computing. Sun Microsystems and others have done work in the grid arena and the technology is still being defined and developed. I do not see where 10gR2 has made any major strides over 10gR1.