Borrow the library to index it, and give it back

The passes that need every photograph's bytes — thumbnails, face
indexing — now borrow each one and release it at the end. On a
placeholder library that is the difference between peak disk being the
working set and being the whole library.

Including on cancellation, which was nearly missed: the face sweep
returns mid-loop when the user presses Stop, and without releasing there
the disk is spent and nothing is delivered for it.

`materialise` now answers whether *it* fetched the content. The pool used
to work that out by listing a file's parent directory — one listing per
file across a library — when the backend already had to `stat` it to
decide whether to ask. One syscall instead of a directory walk, and it
removes the bug class the tests found earlier: a file at the library root
has no `parent()`, so every one of them read as already-downloaded.

**Pinning is the retention control**, and it drives the model the catalog
already had rather than a second one. `tier_desired` is what the user
asked to keep hydrated, `pending_pins` is the resumable work list, and a
pinned collection is never dehydrated for the same reason it was never
evicted. It was in fact *broken* here before: `get` on a stub failed, and
the pin worker logged "one unreadable file must not abandon the whole
pin" and silently did nothing.

Pinned originals on such a library are recorded with `path = NULL`
(`Cache::record_in_place`) rather than copied under `originals/`. Two
reasons, and the second is the important one. A copy would hold every
pinned photograph twice, with the budget able to evict the half that was
not costing the disk. And `release` deletes the file a row names — so a
row that names none cannot delete anything, which puts the one
catastrophic operation out of reach by construction rather than by
remembering not to call it. Deleting a materialised file inside a synced
tree removes the photograph from the server and every other device.

Handing disk back is `spawn_dehydrate`, which asks the client.

Two gaps written down rather than papered over (docs/storage.md §7): a
hydrating pass cannot yet quote its cost, because a stub reports no size;
and the two sweeps hold separate pools, so a library indexed for both
fetches twice.

core/dr-sync-folder/src/borrow.rs

@@ -85,27 +85,6 @@ impl BorrowPool {
         &'p self,
         backend: &dyn RemoteBackend,
         path: &RemotePath,
-    ) -> Result<Borrowed<'p>, RemoteError> {
-        self.borrow_known(backend, path, None).await
-    }
-
-    /// [`borrow`](Self::borrow) where the caller already knows whether the
-    /// content is here.
-    ///
-    /// A sweep reads availability out of the catalog on its way to building a
-    /// work list, so making the pool re-derive it costs a directory listing
-    /// per file for an answer already in hand.
-    ///
-    /// **`None` means "find out", and being wrong is not symmetric.** Claiming
-    /// a file was already local leaves it downloaded — disk, and nothing else.
-    /// Claiming it was not releases a file the user may have pinned. So an
-    /// uncertain caller passes `Some(true)` or `None`, never a guess at
-    /// `false`.
-    pub async fn borrow_known<'p>(
-        &'p self,
-        backend: &dyn RemoteBackend,
-        path: &RemotePath,
-        already_local: Option<bool>,
     ) -> Result<Borrowed<'p>, RemoteError> {
         if !backend.capabilities().materialisation.can_materialise() {
             return Ok(Borrowed {
@@ -129,32 +108,19 @@ impl BorrowPool {
             }
         }
 
-        let id = RemoteId::Path(path.clone());
+        // The backend answers whether *it* fetched the content, because it had
+        // to look before deciding. Determining that here instead would cost a
+        // directory listing per file, and getting it wrong in the wrong
+        // direction releases a file the user pinned.
+        let ours = backend.materialise(&RemoteId::Path(path.clone())).await?;
 
-        // What was here *before* we asked. The whole contract rests on this
-        // being read first: after `materialise` there is no way to tell what
-        // we brought from what was already there.
-        let was_local = match already_local {
-            Some(known) => known,
-            None => is_materialised(backend, path).await,
-        };
-
-        if !was_local {
-            backend.materialise(&id).await?;
-        }
-
-        self.lock().insert(
-            path.clone(),
-            Held {
-                borrowers: 1,
-                ours: !was_local,
-            },
-        );
+        self.lock()
+            .insert(path.clone(), Held { borrowers: 1, ours });
 
         Ok(Borrowed {
             pool: Some(self),
             path: path.clone(),
-            hydrated: !was_local,
+            hydrated: ours,
         })
     }
 
@@ -237,28 +203,5 @@ impl Drop for Borrowed<'_> {
     }
 }
 
-/// Whether the backend currently holds this file's content.
-///
-/// Asked by listing its parent, because the trait has no "stat one object" —
-/// and adding one for this would put a method on every backend to serve a case
-/// only one of them has.
-async fn is_materialised(backend: &dyn RemoteBackend, path: &RemotePath) -> bool {
-    // `parent()` is `None` for a file directly under the library root, whose
-    // parent *is* the root — not "no parent". Treating the two alike reported
-    // every top-level photograph as already downloaded, so nothing was ever
-    // hydrated and nothing was ever released.
-    let parent = path.parent().unwrap_or_else(RemotePath::root);
-    match backend.list(&parent, None).await {
-        Ok(entries) => entries
-            .iter()
-            .find(|e| &e.path == path)
-            .map(|e| e.materialised)
-            // Not listed at all: nothing to hydrate, and the caller's own read
-            // will report the miss properly.
-            .unwrap_or(true),
-        Err(_) => true,
-    }
-}
-
 #[cfg(test)]
 mod tests;