CI: cache mining API + bare-clone data across runs

Adds actions/cache@v4 with per-user keys. Restores the most recent
cache via restore-keys prefix match; saves a fresh cache each run.
Cuts down on duplicate GH API calls (rate-limited 5000/hr) and
re-clones for the same repos run-to-run.

Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>

Author: pirate

.github/workflows/mine-and-deploy.yml

@@ -38,6 +38,23 @@ jobs:
         with:
           node-version: "24"
 
+      # Persist per-user mining caches across runs so we don't re-fetch
+      # PR details, commit stats, etc. that we've already pulled before.
+      # Key includes the inputs.user (or 'full' for cron runs) so single-
+      # user dispatches restore that user's cache specifically.
+      - name: Restore mining caches
+        uses: actions/cache@v4
+        with:
+          path: |
+            cache_*/api/
+            cache_*/bare/
+            cache/api/
+            cache/bare/
+          key: stats-cache-v1-${{ inputs.user || 'full' }}-${{ github.run_id }}
+          restore-keys: |
+            stats-cache-v1-${{ inputs.user || 'full' }}-
+            stats-cache-v1-
+
       - name: Install gh CLI
         run: |
           type -p gh >/dev/null || (