s3: note object names should NOT depend on bucket or server (#1826)

Ref getodk/central#1862

Author: alxndrsn

lib/external/s3.js

@@ -116,13 +116,20 @@ const init = (config) => {
 
   const objectNameFor = ({ id, sha }) => {
     // Include blob ID in object name to allow easy correlation with postgres data.
+    //
     // Include blob SHA sum to prevent name collision in case multiple odk-central-
     // backend instances point to the same bucket.  There are a few scenarios where
     // this could happen, e.g.
     //
     //   * instance reset after testing/training
     //   * staging & prod instances pointed to the same bucket
     //   * temporary loss of access to postgres data on upgrade error
+    //
+    // To allow restoring from backups, do NOT include:
+    //
+    //   * S3 server URL
+    //   * bucket name
+
     if (typeof id !== 'number') throw new Error('Invalid id: ' + id);
     if (!sha) throw new Error('Missing sha sum for blob: ' + id);
     return `${objectPrefix??''}blob-${id}-${sha}`;