Add Retrieve command and integrate database retrieval into screenshot prompts

[Android-PowerUser]

Motivation

• Introduce a retrieve(...) command so the AI can request information stored in the local system-message database and include that content in subsequent screenshot analysis.
• Prevent Retrieve commands from being executed as accessibility actions and instead resolve them into prompt content to enrich AI reasoning for the next screenshot.

Description

• Added a new Command.Retrieve type and recognition for retrieve("...") in CommandParser and updated logging in logCommandDetails and command parsing tests to cover it.
• Implemented retrieval helper functions in PhotoReasoningTextPolicies to format database entries, resolve a retrieval request (resolveRetrievalRequest), format results for prompts, and detect if a heading was already retrieved in chat.
• Updated PhotoReasoningViewModel to filter out Retrieve commands from direct execution, build pendingRetrievedInfoForNextScreenshot when a takeScreenshot() is present, and merge that retrieved info into screenInfo via buildEnrichedScreenInfo when adding screenshots or sending prompts; executeAccessibilityCommand remains unchanged for actionable commands.
• Minor UI and UX changes: show Retrieve in the detected commands list, adjust the database popup explanatory text, change exported filename to Database.txt, and add a `retrieve(

---

Codex Task

[amazon-q-developer]

The PR successfully implements the "Retrieve" command feature for database information retrieval. The implementation is comprehensive and includes:

• New Command.Retrieve class for representing retrieval commands
• Command parser updates to recognize retrieve("heading") syntax
• UI updates to display retrieval commands in the detected commands list
• Text policy functions for resolving retrieval requests and formatting results
• ViewModel integration to enrich screenshot context with retrieved information
• Proper filtering to prevent duplicate retrievals
• Test coverage for the new command parsing

All changes appear to function correctly and follow the existing code patterns. No blocking defects identified.

---

You can now have the agent implement changes and create commits directly on your pull request's source branch. Simply comment with /q followed by your request in natural language to ask the agent to make changes.