harden the ephemeral-root salvage: clone, check space, lock

- clone each file with FICLONE, copy the data only where that fails
- skip the salvage when the copy would leave too little free space
- hold a per-container flock and one copy at a time across containers
- run WriteNspawnFlags on the blocking pool

hive-priv/src/container_config.rs

@@ -61,6 +61,32 @@ pub(crate) fn handle_write_nspawn_flags(
     Ok((String::new(), String::new()))
 }
 
+/// [`handle_write_nspawn_flags`] on the blocking pool. The root salvage inside
+/// it copies files for as long as the root is big, and would hold a runtime
+/// worker for that time.
+///
+/// # Errors
+/// As [`handle_write_nspawn_flags`], or when the task fails to join.
+pub(crate) async fn handle_write_nspawn_flags_blocking(
+    container: String,
+    binds: Vec<BindMount>,
+    isolation: NetworkIsolation,
+    load_credentials: Vec<CredentialMount>,
+    hive_facts: HiveFacts,
+) -> Result<(String, String)> {
+    tokio::task::spawn_blocking(move || {
+        handle_write_nspawn_flags(
+            &container,
+            &binds,
+            &isolation,
+            &load_credentials,
+            &hive_facts,
+        )
+    })
+    .await
+    .context("write nspawn flags task")?
+}
+
 /// Validate a system-level container name (`h-` prefix for all agents,
 /// the manager included) and return the logical agent name after it.
 /// Anything else is refused, the swarm's forge and matrix containers