fix: use triggerAutoBackfill for targeted resync, keep onResync for bulk

Single-repo resync now uses triggerAutoBackfill directly instead of
triggering a full network re-enumeration and process restart. Bulk
resync-all still uses onResync callback in production for memory
reclamation restart.

Co-Authored-By: Claude Opus 4.6 <[e-mail]>

packages/hatk/package.json

···
 {
   "name": "@hatk/hatk",
-  "version": "0.0.1-alpha.36",
+  "version": "0.0.1-alpha.37",
   "license": "MIT",
   "bin": {
     "hatk": "dist/cli.js"

packages/hatk/src/server.ts

···
         } else {
           repoList = await listActiveRepoDids()
         }
+        const isTargeted = Array.isArray(dids) && dids.length > 0
         for (const did of repoList) {
           await setRepoStatus(did, 'pending')
         }
-        if (config.onResync) {
+        if (isTargeted) {
+          for (const did of repoList) {
+            triggerAutoBackfill(did)
+          }
+        } else if (config.onResync) {
           config.onResync()
         } else {
           for (const did of repoList) {