fix: Fixed all 4 bugs from issue #243: initialized tool_calls before

[singlaamitesh]

Fixes #243

Summary

Fixed all 4 bugs from issue #243: initialized tool_calls before the if/else and replaced in-place list.remove() with a filter comprehension to avoid UnboundLocalError and skip-while-iterating in retriever_rag.py; reassigned the discarded str.replace() result so tool descriptions are actually cleaned; moved the streaming yield inside the if msg: guard so None is never sent to clients; and bounded the in-memory chat_history dict with an OrderedDict + eviction helper capped by MAX_IN_MEMORY_CONVERSATIONS (default 1000) to prevent unbounded memory growth when USE_DB=false. Reproduced bug 1's crash with a minimal repro before fixing, and added regression tests for all four bugs.

Changes

• backend/src/agents/retriever_rag.py
• backend/src/api/routers/conversations.py
• backend/tests/test_retriever_graph_simplified.py
• backend/tests/test_api_conversations_streaming.py

How this was tested

Ran the affected tests locally.

[Copilot]

Copilot was unable to review this pull request because the user who requested the review has reached their quota limit.

[luarss]

Thanks @singlaamitesh , salvaged in #311