The Playwright screenshot path waited for `.loading` elements to detach
by calling `page.locator(".loading").all()` once and iterating the
snapshot. Spinners that appeared after the snapshot was taken (common on
long dashboards where charts load lazily) were never waited for, causing
PDF reports to capture charts that were still fetching data.
Replace the snapshot loop with `page.wait_for_function` which
continuously polls `document.querySelectorAll('.loading').length === 0`
until the page is genuinely clear of all spinners, then use
`self._screenshot_load_wait` (already stored on the base class) for the
timeout to stay consistent with the Selenium path.
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
