Currently we have a map which maintains a mapping of CIM & containerd ID to the volume at
which a CIM is mounted for the given container. This was required before the layer
refactoring work when we needed to get the volume path from the layer cim path. However,
this isn't needed anymore. As of now, this map doesn't provide much value and makes the code a
bit complicated. Moreover, we will need to rewrite some of this code anyway when we do the work
required for handling `shim delete` cleanups properly (containerd/containerd#9727).

Signed-off-by: Amit Barve <ambarve@microsoft.com>

CimFS now supports a new format for storing CIMs, named BlockCIM.  A block CIM format can
store the entire CIM on a block device (like a VHD) or a file formatted like a block
device.

This commit adds Go wrappers for the new CimFS APIs that allow creation, merging and
mounting of such Block CIMs. Some new flags required when creating and mounting these CIMs
are added and some deprecated flags have been removed. New type has been introduced to
represent a block CIM. Unit tests have been added to test the newly added CimFS
functionality. Lastly, CimFS flags aren't a part of the hcs schema (only the CimMount
request is), those flags are moved from the hcs/schema2 package to the cimfs package.

Signed-off-by: Amit Barve <ambarve@microsoft.com>

This commit adds a layer writer that can be used for extracting an image layer tar into a
Block CIM format.

Existing forked CIM layer writer was renamed to a common base type `cimLayerWriter`.
Forked CIM layer writer & Block CIM layer writer both now extend this common base type to
write layers in that specific format.

Signed-off-by: Amit Barve <ambarve@microsoft.com>

This commit adds the ability to parse block CIM layer mounts and to mount the merged block
CIMs to be used as a rootfs for a container.

Signed-off-by: Amit Barve <ambarve@microsoft.com>