Fix what review found in the Disk section

Safety:

- `destroy`'s rollback-pin arm took a tag over IPC and interpolated it
  straight into an image reference it then removed. `tag: "latest"` named
  the project's live snapshot, deleted under a dialog saying "rollback
  pin". It is the one destructive variant carrying a free-form string, so
  it now goes through `parse_rollback_tag`.
- The compaction's scratch container was named `triple-c-scrub-*`, which
  is what the scrub reclaim bucket hunts and force-removes. A reclaim from
  a second window would have destroyed the container a running compaction
  was about to commit. It gets `triple-c-compact-*`, swept at the start of
  the next compaction rather than from a bucket anything else can fire.
- Deleting a home or config volume only refused a *running* container, but
  a stopped one still pins its volumes — the resting state of every
  project ever started — so the user typed the project name and met a raw
  409. The container is now removed first and `loses` says so.

Correctness:

- The compaction Dockerfile emitted no `LABEL`, so the flattened
  intermediate could never match the sweep's `dangling` + `triple-c.managed`
  filter that three cleanup paths rely on. Verified on Docker 29.7.2 that
  the label lands on the final stage, the build still yields one layer, and
  untagging the staging tag after the commit leaves the committed snapshot
  intact and startable.
- `snapshot_commit_layers` silently meant something else when
  `triple-c.base-image-id` was absent — the normal case for a pre-label
  project — counting the base's own layers and letting a never-recreated
  project qualify for compaction. `base_lineage_known` now carries that,
  the column says "unknown", and the plan does not offer the rewrite.
- `destroy` returned a `ReclaimResult` wearing a `ReclaimTarget` that named
  work it had not done (a home-volume deletion came back as
  `OrphanVolume`). Split into `target` / `destroyed`, exactly one set.
- `formatBytes` ran `toFixed` after the divide loop, so 999,999 rendered as
  "1000.0 KB" — in the app's only byte formatter, in a panel full of
  near-boundary sizes.
- `is_base_image_reference` split on the first colon, so a registry port
  ate the repo name.

UI:

- `snapshot_above_base_bytes: null` — deliberately unmeasurable — rendered
  as "0 B", the one guessed number in the table.
- Layer count was flagged by colour alone; it now says "stacked".
- The tick list survived a reclaim, so the same call could be re-fired at
  objects that no longer existed. The plan is dropped after any action and
  the panel says the totals predate it.
- `setReport` landed before the plan call was awaited, so a plan failure
  rendered fresh totals above the previous scan's rows.
- Both confirmation modals unmounted before awaiting, making the entire
  busy path dead code during multi-second work.
- `buildx du` failures silently showed `docker system df`'s under-reported
  build-cache figure with no explanation.
- Tooltip text reached no assistive tech, so two headers announced as
  "Help"; hardcoded input id; error-toned glyph in warning-toned panels;
  `sweepOrphanedSnapshots` and `clearOutcome` had no callers.
- Four docstrings claimed things the code did not do, and two tests were
  named for behaviour they did not assert.

Tests: 513 frontend (was 502), 370 Rust (was 365).

Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01GBq2rGum6GX7xXgsas1fDc

app/src/lib/formatBytes.ts

@@ -1,10 +1,14 @@
 /**
  * The one byte formatter.
  *
- * Before this existed the app had four of them — `projects/home/format.ts`,
+ * The app had four of them — `projects/home/format.ts`,
  * `projects/migrationCopy.ts`, `settings/UpdateDialog.tsx` and an inline
  * `toFixed(1)` in `useProjectActions.ts` — disagreeing about the divisor, the
- * unit labels and the precision. They are now expressed in terms of this.
+ * unit labels and the precision. The first two now delegate here.
+ *
+ * The other two deliberately do not, yet: `UpdateDialog` renders KB at
+ * `toFixed(0)`, so re-pointing it would change what a download size reads as,
+ * and neither is on the Disk panel's path. They are the remaining copies.
  *
  * ## Why the default is base 1000
  *
@@ -14,8 +18,12 @@
  * same build cache would read as a bug in the panel. So decimal is the default
  * and binary is opt-in, rather than the other way round.
  *
- * Both existing conventions are preserved exactly, so re-pointing the old
- * call sites changed no rendered string:
+ * Both existing conventions are preserved for every size either call site can
+ * realistically produce — a file size or a payload size, i.e. a non-negative
+ * finite number below a terabyte. Outside that range this deliberately differs
+ * from what it replaced: a negative or `NaN` input now renders `—` rather than
+ * `-1 B` or `NaN GB`, and the unit ladder continues past GB instead of
+ * stopping there.
  *
  * - `{ }`                        → `41.0 MB`   (decimal, what migration used)
  * - `{ binary: true }`           → `1.5 GB`    (÷1024 with decimal-style
@@ -56,6 +64,17 @@ export function formatBytes(bytes: number, options: FormatBytesOptions = {}): st
     value /= step;
     unit += 1;
   }
+
+  // **Promote again if rounding pushed the value back up to a whole step.**
+  // `toFixed` runs after the loop, so 999,999 B divides to 999.999 KB and then
+  // renders as "1000.0 KB" — a unit the loop had already decided against. The
+  // same happens at every boundary (999,999,999 → "1000.0 MB", and 1,048,575
+  // → "1024.0 KB" in binary).
+  if (unit < units.length - 1 && Number(value.toFixed(precision)) >= step) {
+    value /= step;
+    unit += 1;
+  }
+
   // Whole bytes never get a decimal point: `512 B`, not `512.0 B`.
   return unit === 0
     ? `${Math.round(bytes)} ${units[0]}`
@@ -73,7 +92,7 @@ export function formatBytesDelta(bytes: number, options?: FormatBytesOptions): s
 }
 
 /**
- * `up to 12.3 GB` / `nothing` — for a bound rather than a measurement.
+ * `up to 12.3 GB` — for a bound rather than a measurement.
  *
  * The Disk panel is careful about this distinction: every figure it shows is
  * measured except a compaction's yield, which cannot be known until it runs.