fix(cli): ignore closed readline after stdin (#8862)

What changed

Wrap the stdin readline async iterator so CSpell ignores the terminal readline was closed error that can occur after stdin has already yielded its file list.

Other readline errors are still rethrown.

Why

Fixes #8778. On Node 24, cspell lint --file-list stdin can finish processing a large stdin file list and then report Linter Error: readline was closed instead of completing cleanly.

Tests

Added unit coverage for:

• preserving all yielded stdin lines when readline throws readline was closed after reading
• continuing to throw unrelated readline errors

I could not run the package test command locally in this sandbox because the checkout has no node_modules, and Corepack is blocked from creating its user cache directory (EPERM under ~/.cache/node/corepack).

fix(docs): replace .json with package.json in configuration docs (#8861)

Fix

In the package.json configuration section, the sentence referred to
"the .json file" instead of "the package.json file".

Before: "CSpell looks for the configuration in the cspell field of the .json file."
After: "CSpell looks for the configuration in the cspell field of the package.json file."

fix: substitution display and ranges for output checking (#8838)

Added a French dictionary and fixed the positioning logic for error reporting of replacement words.

fix: use env vars for GitHub context expressions in run steps to prevent injection (#8852)

Summary

This PR fixes script injection and secret exposure security issues found in GitHub Actions workflows and composite actions. In each case, a GitHub expression (${{ ... }}) was used directly inside a run: shell command instead of being passed through an environment variable — the recommended approach per GitHub's security hardening guide.

Issues Fixed

🔴 Secrets exposed in process arguments (visible via ps aux)

🔴 GITHUB_ENV injection (allows setting arbitrary environment variables)

🟠 Shell script injection via unsanitised expressions in run: steps

Fix Pattern

For each issue, the expression is moved to an env: block and referenced via shell variable:

# Before (unsafe)
run: some-command ${{ github.event.inputs.ref }}

# After (safe)
env:
  REF: ${{ github.event.inputs.ref }}
run: some-command "$REF"

For jq queries using matrix.repo, the injection-safe --arg flag is used:

# Before (jq injection risk)
run: jq '.[] | select(.path == "${{ matrix.repo }}")'

# After (safe)
env:
  REPO: ${{ matrix.repo }}
run: jq --arg repo "$REPO" '.[] | select(.path == $repo)'