fix(templates): Pin playwright to base image version in uv Dockerfile template

[vdusek]

Summary

The uv package-manager branch of the actor Dockerfile template (src/crawlee/project_template/{{cookiecutter.project_name}}/Dockerfile) doesn't pin Python playwright to the version pre-installed in the Apify base image. When the lockfile resolves to a different playwright version than the base image (currently 1.59.0 vs the base image's 1.60.0 + chromium-1223), the actor tries to launch a Chromium build that /pw-browsers/ doesn't ship and fails with Executable doesn't exist at /pw-browsers/chromium_headless_shell-1217/....

The pip and poetry branches solve this by sed-rewriting playwright==X.Y.Z to whatever playwright --version reports in the image. The uv branch's --no-install-package playwright was supposed to be the equivalent, but the test's pip install crawlee --force-reinstall (and any user step that touches playwright before the uv sync runs) defeats it.

This fix mirrors the pip/poetry intent for uv:

• Snapshot the base image's playwright --version in the early pip install uv layer, before COPY pyproject.toml uv.lock ./ or any test-injected step can mutate it.
• After uv sync ... --no-install-package playwright, check if the env's playwright still matches; if it has drifted, force-reinstall with uv pip install --reinstall --no-deps "playwright==<base-version>".
• Guard the reinstall behind a version check so the common case (no drift) skips the wheel download.

Resolves the 9 failing uv-variant scheduled e2e jobs in run 26082797876 (playwright, adaptive_beautifulsoup, adaptive_parsel × {httpx, curl_impersonate, impit}).

Verified against the live apify/actor-python-playwright:3.13 image:

Scenario	Before	After
Real user, no drift	works, but pip freeze | grep playwright check is brittle	guard short-circuits, no extra work
Test injection downgrades playwright to 1.59.0	mismatched chromium → crash	reinstall pins to 1.60.0, browsers match

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 92.89%. Comparing base (7edfe0d) to head (709d61a).

Additional details and impacted files

@@            Coverage Diff             @@
##           master    #1904      +/-   ##
==========================================
+ Coverage   92.86%   92.89%   +0.02%     
==========================================
  Files         167      167              
  Lines       11709    11709              
==========================================
+ Hits        10874    10877       +3     
+ Misses        835      832       -3     

Flag	Coverage Δ
unit	92.89% <ø> (+0.02%)	⬆️

Flags with carried forward coverage won't be shown. Click here to find out more.

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.
• 📦 JS Bundle Analysis: Save yourself from yourself by tracking and limiting bundle sizes in JS merges.

[janbuchar]

If we do this check every time, what is the point of the --no-install-package playwright? It looks like a belt-and-suspenders thing to me.