[wip] feat: docs ai - Conversational search

[onceLearner]

What does this PR do? What is the motivation?

⚠️ ADDRESS TODO

Adds Docs AI [ Conversational Search] to the docs site, gated behind a feature flag.

Core implementation

• New UI templates: layouts/partials/conversational-search.html (dialog shell, empty state, message actions).

• New JS module: assets/scripts/components/conversational-search.js to render/clone templates, manage chat lifecycle (new question/reset), message actions (copy + thumbs), resources/source chips, and link formatting.

• Instantsearch integration:

  • Adds AI suggestion list above hits and includes it first in keyboard navigation.
  • Updates suggestion text live from the input (avoids debounce lag/flicker).
  • Handles Enter on the AI row to open the assistant and closes the dropdown.

• Feature flags via OpenFeature + Datadog provider:

  • Adds assets/scripts/helpers/feature-flags.js with a singleton init.
  • Uses RUM-derived targetingKey (device/session) for weighted rollouts; temporarily blocks until RUM context is available.

• Observability:

  • Emits structured Logs + RUM actions for key interactions (suggestion click, copy, feedback, etc.).
  • Enables RUM experimental feature_flags for tracking.

• Build/dependency updates:

  • Upgrades marked to v17 and switches import usage to the UMD build for CommonJS scripts.
  • Adds highlight.js + marked-highlight.
  • Bumps Typesense client to v3 and adapter to v2.9.
  • Removes dead Algolia config and excludes embedding from Typesense payloads.

  User-facing entry points

• “Ask AI” button (global) that opens a modal sidebar chat.

• Homepage hero CTA (“Ask AI” beside search) when the flag is enabled.

• Search dropdown: injects an AI suggestion row (“Ask AI about ‘’”) that is keyboard-navigable and triggers the chat on Enter/click.

Motivation
Ship a controlled Docs AI preview to improve discovery and reduce search friction, while measuring adoption/latency/click-through via RUM/logs and rolling out safely with flags.

---

Merge instructions

Merge readiness:

• Ready for merge

Notes before merging:

• PR is Draft and branch is out-of-date with master → rebase/merge master before final merge.
• Requires ≥1 approving review from maintainers with write access.
• If Doggo/Vault fails, confirm AppGate/Vault access and re-run checks.

---

Additional notes

Commit highlights (high level):

• Initialization + UI buildout (templates + sidebar dialog).
• Searchbar + dropdown AI suggestion, including keyboard support.
• Feature flag singleton + removal of top-level await issues.
• Logging/RUM instrumentation (including copy + click tracking).
• Dependency bumps (marked v17/UMD, Typesense v3), security hardening (XSS fix), and payload trimming (embedding exclusion).

[github-actions]

Preview links (active after the build_preview check completes)

New or renamed files

• https://docs-staging.datadoghq.com/reda.elissati/docseng-249-docs-ai-preview-rollout/product_analytics/session_replay/mobile/privacy_options
• https://docs-staging.datadoghq.com/reda.elissati/docseng-249-docs-ai-preview-rollout/product_analytics/session_replay/mobile/setup_and_configuration
• https://docs-staging.datadoghq.com/reda.elissati/docseng-249-docs-ai-preview-rollout/real_user_monitoring/session_replay/mobile/privacy_options
• https://docs-staging.datadoghq.com/reda.elissati/docseng-249-docs-ai-preview-rollout/real_user_monitoring/session_replay/mobile/setup_and_configuration