feat(ai): add conversation memory to Ask ApplyWell

[vitorhugo-dotnet]

Summary

Implements #93 using Spring AI native chat memory so Ask ApplyWell can resolve follow-up questions from the active conversation context.

Implementation

• adds required conversationId: UUID to AssistantChatRequest
• wires MessageChatMemoryAdvisor with MessageWindowChatMemory
• supplies ChatMemory.CONVERSATION_ID on every assistant request
• namespaces conversation memory server-side as <userId>:<conversationId> to prevent cross-user leakage
• bounds conversation memory to 20 messages by default via app.assistant.memory-max-messages
• updates the system prompt to preserve relevant organization/application/recruiter/status/platform/date-range filters across follow-ups
• keeps existing read-only assistant tools unchanged
• keeps the SSE event contract unchanged (token, error, complete)
• does not introduce vector DB, embeddings, RAG, or persistent chat storage

Verification

TDD cycle was exercised explicitly:

• RED: 425 tests, 6 expected failures from the new conversation-memory specification
• GREEN: conversation memory implementation added
• regression investigation: one test incorrectly inspected a record-component annotation; replaced with real Jakarta Validation behavior
• final mvn -B -ntp clean verify: 425 tests, 0 failures, 0 errors, BUILD SUCCESS
• AssistantConversationMemoryTest: 6/6 passing
• Qodana workflow: success

The Qodana PR comment lists three repository-wide warnings. This workflow uses pr-mode: false, and the same three warnings were already reported on the initial test-only RED commit, so they are not introduced by the production implementation in this PR.

Covered scenarios

• same conversationId retains previous user/assistant turns
• different conversation IDs are isolated
• the same client conversation UUID used by different authenticated users is isolated by server-side user namespacing
• request validation requires a conversation ID
• memory window is bounded
• follow-up filter preservation is explicitly included in the system prompt

Persistence

This PR intentionally uses Spring AI's in-memory repository. Persistence across application restarts or multiple backend replicas was optional in #93 and is not required for the current deployment. The ChatMemory abstraction keeps a future move to JdbcChatMemoryRepository isolated from the HTTP contract.

Closes #93

[github-actions]

Qodana for JVM

3 new problems were found

Inspection name	Severity	Problems
Nullability and data flow problems	🔶 Warning	1
Mismatched query and update of 'StringBuilder'	🔶 Warning	1
Simplifiable conditional expression	🔶 Warning	1

View the detailed Qodana report

To be able to view the detailed Qodana report, you can either:

• Register at Qodana Cloud and configure the action
• Use GitHub Code Scanning with Qodana
• Host Qodana report at GitHub Pages
• Inspect and use qodana.sarif.json (see the Qodana SARIF format for details)

To get *.log files or any other Qodana artifacts, run the action with upload-result option set to true,
so that the action will upload the files as the job artifacts:

      - name: 'Qodana Scan'
        uses: JetBrains/qodana-action@v2025.3.2
        with:
          upload-result: true

Contact Qodana team

Contact us at qodana-support@jetbrains.com

• Or via our issue tracker: https://jb.gg/qodana-issue
• Or share your feedback: https://jb.gg/qodana-discussions