chore: bump leanVM to 8fcbd779 and leanSpec to latest main

.github/workflows/ci.yml

@@ -117,10 +117,16 @@ jobs:
           path: leanSpec/packages/testing/src/consensus_testing/test_keys/prod_scheme
           key: ${{ steps.cache-prod-keys.outputs.cache-primary-key }}
 
+      # `-n 2` (not `-n auto`) caps the number of concurrent leanVM provers.
+      # The devnet5 prover peaks at ~4-5 GiB per proof; `-n auto` on the
+      # 4-vCPU/16 GiB GitHub runner spawns 4 provers (~18 GiB) and OOM-kills
+      # the runner mid-generation. Proving is core-bound (rayon saturates all
+      # cores per proof), so fewer workers barely affects wall-time. The
+      # Makefile keeps `-n auto` for local machines with more memory.
       - name: Generate test fixtures
         if: steps.cache-fixtures.outputs.cache-hit != 'true'
         working-directory: leanSpec
-        run: uv run fill --fork Lstar -n auto --scheme prod -o fixtures
+        run: uv run fill --fork Lstar -n 2 --scheme prod -o fixtures
 
       # Save fixtures even if a later step fails, so a re-run does not
       # have to regenerate them. See: https://github.com/actions/cache/tree/main/save#always-save-cache