CI: Fix recurring Rust toolchain flake in the libjs-test262 job

Problem: The libjs-test262 job was failing — often — with “cargo is not
installed for the toolchain 1.96.0”, while building the Rust crates.

Cause: The job ran in an older CI image lacking 1.96.0 — making runtime
rustup try to auto-install 1.96.0 the first time a proxied cargo ran.
And because the build compiles multiple Rust crates in parallel, several
cargo invocations hit the missing toolchain at once — and each started a
rustup install of the same toolchain. And rustup apparently isn’t safe
against concurrent installs: Its in-place component swap shuffles files
out to a temp directory, and a crate that reads the toolchain mid-swap
sees rustc — but not cargo. Whether it corrupts depends on timing — so
the failure is intermittent, and a re-run usually “fixes” it.

Fix: Update the job to the current ladybird-ci:2026.06.02 image — which
already has 1.96.0 (with cargo) baked in — so no runtime install happens
and there’s nothing to race. As an additional guard, also install the
pinned toolchain *explicitly*, before the parallel build: so any install
happens exactly once — and serially — rather than being raced.

.github/workflows/libjs-test262.yml

@@ -19,7 +19,7 @@ jobs:
     runs-on: 'blacksmith-16vcpu-ubuntu-2404'
     if: github.repository == 'LadybirdBrowser/ladybird'
     container:
-      image: ghcr.io/ladybirdbrowser/ladybird-ci:2026.05.25
+      image: ghcr.io/ladybirdbrowser/ladybird-ci:2026.06.02
 
     steps:
       - name: Checkout LadybirdBrowser/ladybird
@@ -79,6 +79,9 @@ jobs:
           mkdir -p old-libjs-data
           cp -R libjs-data/test262 libjs-data/wasm old-libjs-data
 
+      - name: Warm up the Rust toolchain
+        run: rustup toolchain install
+
       - name: Build test262-runner, test-js, test-wasm and test-js-bytecode
         run: |
           cmake --preset Release -B libjs-test262/Build \