Description of problem:
qemu-img-[[rh]ev]-2.10 provided the ability to perform "unsafe" operations on the leaf of a qcow2 chain, without preparing the entire backing chain (see bug 1213786) for details.

This BZ is a tracker for mapping out places where vdsm needlessly prepares the entire chain and fixing them so they prepare the minimal number of volumes, if at all.
Fixing these would both improve performance and reduce the risk of leaving "leftover" prepared volumes they may later cause problems.

Version-Release number of selected component (if applicable):
vdsm-4.20

An initial list of flows to consider:
- Creating a snapshot (already handled by bug 1395941).
- Creating a thin VM/pool on top of a snapshot (possibly handled by the previous item)
- Merging a snapshot (only the two volumes merged should be activated, at least in theory)
- Amending qcow compat level
- Extending the size of a disk
- getVolumeInfo
-