Add GPT partition identity to ZInfoStorage

[eriknordmark]

Summary

Extend ZInfoStorage (the per-partition entry in
ZInfoDevice.storageList) with three optional fields so a controller
can identify each entry by its GPT role rather than only by raw device
name:

• partitionLabel (5) — GPT PARTLABEL, e.g. EFI System, CONFIG,
  IMGA, IMGB, P3.
• partitionTypeGuid (6) — GPT partition-type GUID.
• partitionUuid (7) — GPT PARTUUID — the same identifier EVE's boot
  path uses to resolve IMGA / IMGB / CONFIG.

Motivation

EVE already enumerates every block partition lsblk sees and reports
each one's size in ZInfoStorage.total. But a controller seeing
/dev/sda1, /dev/sda2, /dev/sda3 with sizes cannot tell which is
the EFI partition, which is IMGA vs IMGB, or compare the size of
either against the rootfs being staged. With these three fields it
can.

Compatibility

Pure additive change. New fields are appended as 5/6/7; existing
clients keep parsing the message unchanged. EVE fills the fields only
for block-partition entries on GPT-partitioned disks; directory
entries and partitions on non-GPT disks emit empty strings.

Follow-up

EVE pillar needs a follow-up change to populate the new fields from
lsblk -o PARTLABEL,PARTTYPE,PARTUUID (or equivalent). That change
is not in this PR.

[rene]

This comment is a bit odd. Device carrying GPT identification will be block devices like sda, sdb, etc, not sda1, sda2, sda3 which represents the partition itself.... and why device != "". Please, review the comment...

[eriknordmark]

The device field can be either /dev/sda or /dev/sdaN and all of those are block devices in Linux.
The comment refers to the underying disk (aka /dev/sda) being GPT.
The GPT info is contained in /dev/sda but is associated with the partitions.

So I don't see anything wrong in the comments - maybe this can be made more clear somehow. Let me try.

[eriknordmark]

You're right — "block-partition" lumped whole disks and partitions together, and device != "" is the wrong gate (the empty-device case is the directory entries from ReportDirPaths, not whole-disk entries).

Rewrote the comments in the latest force-push: dropped the umbrella note and made each field's comment state its own emptiness rules — the GPT fields are set only when the entry is a GPT partition, and empty for whole-disk entries (e.g. sda), for partitions on non-GPT disks, and for directory entries.