Use completed() command for automatic screenshot flow

[Android-PowerUser]

Motivation

• Replace the old explicit takeScreenshot() requirement with a completed() signal so the AI can stop automatic screenshot requests when the task is finished.
• Update the system prompt and runtime flow to make screenshots the default after each AI reply unless completed() is present.

Description

• Add Command.Completed to the command sealed class and make completed() detectable by the parser and single-instance guarded in CommandParser.
• Extend PhotoReasoningCommandProcessing to report hasCompletedCommand, and update PhotoReasoningViewModel to automatically append a TakeScreenshot for final execution when neither completed() nor an explicit takeScreenshot() is present, while treating completed() as a termination marker (no accessibility action).
• Stop incremental command execution when completed() is encountered during streaming and do not dispatch it to accessibility (no-op in ScreenOperatorAccessibilityService).
• Update default system message in SystemMessagePreferences to instruct use of completed() and to reflect the new behavior "After each message... Say completed() when the task is finished.".
• Surface completed() in the detected-commands UI and add unit tests to cover completed() parsing in CommandParserTest.

Testing

• Ran ./gradlew :app:compileDebugKotlin and the build completed successfully (BUILD SUCCESSFUL).
• Added unit tests in app/src/test/java/com/google/ai/sample/util/CommandParserTest.kt for completed() parsing; these tests were added but not executed in this compile step.

---

Codex Task

[amazon-q-developer]

The implementation of the completed() command is well-structured and comprehensive. All changes correctly integrate the new command across the application stack - from command parsing and execution to UI display and testing. The code functions correctly with no defects found.

---

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.