setup_builder: wait for non-zero device size before formatting sticky disk

[piob-io]

Summary

Fixes a hot-attach race where mkfs.ext4 fails with Device size reported to be zero right after the sticky-disk response arrives (seen in the staging benchmark run: job thrash-swarm-shared / round2 (pyjobs).

The VM agent hydrates the sticky disk into a pre-attached drive and asks the hypervisor to raise a virtio config-change interrupt; the GetStickyDisk response can reach the runner before the guest kernel has processed that interrupt, so /dev/vdb transiently reports size 0. The same window exists on the direct rbd-nbd hot-patch path, so the fix is guest-side:

setupStickyDisk():
  device = getStickyDisk().device
+ waitForNonZeroDeviceSize(device, 10_000)   // poll `blockdev --getsize64` every 100ms
  maybeFormatBlockDevice(device)             // blkid / resize2fs / mkfs.ext4
  mount(device, mountPoint)

Beyond the mkfs failure, formatting/probing a zero-size view could also misread a warm disk as unformatted, so gating all device access on a non-zero size is the safe ordering. Times out loudly after 10s (observed race window is sub-second). dist/ rebuilt.

Link to Devin session: https://app.devin.ai/sessions/ba61e842e479410d95116fe1dfe77b0f
Requested by: @piob-io

---

^{Need help on this PR? Tag /codesmith with what you need. Autofix is disabled.}

---

^{Need help on this PR? Tag /codesmith with what you need. Autofix is disabled. (Staging)}

[devin-ai-integration]

🤖 Devin AI Engineer

I'll be helping with this pull request! Here's what you should know:

✅ I will automatically:

• Address comments on this PR. Add '(aside)' to your comment to have me ignore it.
• Look at CI failures and help fix them

Note: I can only respond to comments from users who have write access to this repository.

⚙️ Control Options:

• Disable automatic comment, CI, and merge conflict monitoring