test: de-flake cart-drawer overflow spec via consent pre-seed

[satra]

Summary

De-flakes site/src/tests/e2e/cart_drawer_overflow.spec.ts. The test dismissed the first-visit analytics-consent banner reactively — clicking "decline" only if the banner happened to be visible at the moment it checked. On a fresh context the banner can render a beat after that check, stay up, overlay the cart drawer's footer buttons, and intercept pointer events, failing the test intermittently.

This pre-seeds the consent choice in localStorage via page.addInitScript before navigation, so the banner never renders. Deterministic, simpler, and faster.

Scope

• One e2e test file only — no product-code change, no change to the deployed site (byte-identical build).
• Unrelated to the year-slider work (feat(025): NeuroScape atlas year range slider (dual-handle + draggable window) #67) and to the separate preview-infra flakiness tracked in Flaky: ios-siblings WebKit "settled" check times out on PR-preview e2e #68.

Verification

• Added lines pass prettier (the one remaining prettier warning on this file is a pre-existing long line I did not touch).
• The behavioural check is the e2e run itself against the preview.

🤖 Generated with Claude Code

[gemini-code-assist]

Code Review

This pull request refactors the cart_drawer_overflow.spec.ts end-to-end test to pre-set the analytics consent choice to 'denied' in localStorage using page.addInitScript before navigating. This replaces the previous logic that attempted to dynamically find and click the decline button on the consent banner. The reviewer suggested also setting window.__ohbmAnalyticsConsent = 'denied' in the init script to ensure the banner is successfully blocked even in environments where localStorage is disabled or throws an error.

Important

The consumer version of Gemini Code Assist on GitHub is being sunset. Starting June 18, 2026, new organization installations will be blocked, and all code review activity will officially cease on July 17, 2026.
For more details on the timeline and next steps, please review the Help Documentation.

[gemini-code-assist]

To make the consent suppression even more robust, we can also set window.__ohbmAnalyticsConsent = 'denied' directly in the init script.

Since ConsentBanner.svelte checks window.__ohbmAnalyticsConsent on mount and suppresses the banner if it is set to 'denied', doing this ensures the banner is successfully blocked even in test environments where localStorage is disabled or throws an error.

Suggested change

	await page.addInitScript(() => {
	try {
	localStorage.setItem('ohbm2026.analytics.consent.v1', 'denied');
	} catch {
	/* storage unavailable — banner will show; not this test's concern */
	}
	});
	await page.addInitScript(() => {
	try {
	localStorage.setItem('ohbm2026.analytics.consent.v1', 'denied');
	} catch {
	/* storage unavailable — banner will show; not this test's concern */
	}
	(window as any).__ohbmAnalyticsConsent = 'denied';
	});

[satra]

Thanks @gemini-code-assist — I checked the bootstrap and am not taking this suggestion, because setting the global in addInitScript is a no-op here:

• src/app.html's inline consent script runs on every load after Playwright's addInitScript, and unconditionally reassigns window.__ohbmAnalyticsConsent = state (app.html:96). It never reads a pre-existing global — it recomputes state from host / DNT / GPC / localStorage (:87-91). So a global set in the init script is overwritten before ConsentBanner mounts.
• The correct lever is exactly the localStorage pre-seed this PR adds: the bootstrap reads ohbm2026.analytics.consent.v1 at app.html:88 and maps 'denied' → suppressed banner.
• The localStorage-disabled edge doesn't benefit either — the global would still be clobbered by :96, and that path resolves via the bootstrap's own fallbacks, not the test.

(For transparency: on the CI preview host the path is /pr-<N>/…, which short-circuits to 'pr-preview' at app.html:74-76 and suppresses the banner regardless — so this change is a robustness fix for local-preview / production first-visit runs, not the CI flake. Keeping the diff minimal and localStorage-only.)

[gemini-code-assist]

Thanks for the detailed explanation, @satra. That makes perfect sense. I appreciate the clarification on how the app.html logic interacts with the addInitScript and why the localStorage pre-seed is the correct, robust approach for handling this in your test environment. Your reasoning for keeping the diff minimal and focused on the test's stability is sound.