fix: separate CI cache volumes to prevent concurrent disk exhaustion

Each workflow job now uses its own cache directory to avoid multiple
runners on the same VPS writing to the same target dir simultaneously.

.github/workflows/docs.yml

@@ -26,7 +26,7 @@ jobs:
     container:
       image: ${{ needs.preflight.outputs.IMAGE }}
       volumes:
-        - /cache/cargo-target/pezkuwi-sdk:/cache/target
+        - /cache/cargo-target/pezkuwi-sdk-test-doc:/cache/target
     env:
       RUSTFLAGS: "-Cdebug-assertions=y -Dwarnings"
       SKIP_WASM_BUILD: 1
@@ -52,7 +52,7 @@ jobs:
     container:
       image: ${{ needs.preflight.outputs.IMAGE }}
       volumes:
-        - /cache/cargo-target/pezkuwi-sdk:/cache/target
+        - /cache/cargo-target/pezkuwi-sdk-rustdoc:/cache/target
     env:
       SKIP_WASM_BUILD: 1
       CARGO_INCREMENTAL: 0