Optimize noraneko-runtime checkout with partial clone

[Copilot]

The package workflow clones the full Firefox source tree from noraneko-runtime, but artifact mode only needs a subset of files. Full clone is wasteful.

Changes

• Added fetch-depth: 1 for shallow clone (no git history)
• Added filter: blob:none for blobless partial clone (on-demand blob fetching)

- name: Checkout ${{ github.repository }}-runtime
  uses: actions/checkout@v4
  with:
    repository: ${{ github.repository }}-runtime
    ref: ${{ steps.get-commit.outputs.commit_hash }}
    submodules: "recursive"
    fetch-depth: 1
    filter: blob:none

With blob:none, git fetches only tree/commit objects initially; blobs are downloaded on-demand when accessed. Since artifact builds don't compile from source, most source blobs are never fetched.

Original prompt

Check github workflows, and find optimal way to reduce repository size with partial cloning from noraneko-runtime. the noraneko-runtime have full Firefox source code, but in noraneko final package, it uses few scripts from the repository, so full-clone seems waste of time. so, research how to optimize the workflow.

---

💬 We'd love your input! Share your thoughts on Copilot coding agent in our 2 minute survey.