fix(preprocessor): drop dead choice.stop_reason assignment

[krishung5]

Summary

Main HEAD currently fails to compile:

```
error[E0609]: no field `stop_reason` on type `&mut dynamo_protocols::types::ChatChoiceStream`
--> lib/llm/src/preprocessor.rs:1581
1581 | choice.stop_reason = None;
| ^^^^^^^^^^^ unknown field
= note: available fields are: `index`, `delta`, `finish_reason`, `logprobs`
```

Cause: two PRs landed today in a way that left the tree internally inconsistent —

• fix: Add parity in behavior and tests with vllm nemotron_v3 for tools parser #9058 (`f6117b82ec`) added `choice.stop_reason = None;` in the eof-flushed-buffers fallback path
• feat(sglang): tokens in/out with logprobs #8119 (`74e3e0ef15`) removed `stop_reason` from `ChatChoiceStream` as part of the sglang logprobs refactor

The next commit on main (#9230 `8cee1e50e1`, "sglang 0.5.11 bump") only touched sglang/Dockerfile files, so the Pre Merge workflow's path filter skipped `rust-clippy`/`rust-tests` entirely. The broken combination landed silently and is now exposed on every downstream PR that triggers a fresh build (e.g. #9352).

Fix

Drop the orphaned assignment. The surrounding code already resets every remaining field on the struct.

Test plan

• `cargo check -p dynamo-llm` — passes
• preprocessor.rs grep `stop_reason` returns no matches
• CI rust-clippy / rust-tests passes

🤖 Generated with Claude Code

---

Summary by CodeRabbit

• Bug Fixes
  • Improved stream cleanup logic to ensure reasoning content is properly cleared without leaving fragments in processed output.

[coderabbitai]

Caution

Review failed

The pull request is closed.

ℹ️ Recent review info

⚙️ Run configuration

Configuration used: Path: .coderabbit.yaml

Review profile: CHILL

Plan: Pro

Run ID: e3865282-de96-4ec5-a7ac-096ad554d0ca

📥 Commits

Reviewing files that changed from the base of the PR and between 8cee1e5 and d86c19f.

📒 Files selected for processing (1)

• lib/llm/src/preprocessor.rs

---

Walkthrough

lib/llm/src/preprocessor.rs modifies the EOF flush logic in strip_leading_reasoning_start_from_stream to clear additional delta fields (refusal and reasoning_content) when emitting buffered undecided initial chunks, preventing partial reasoning fragments from propagating to output.

Changes

Stream Delta Field Cleanup

Layer / File(s)	Summary
EOF Buffer Flush with Delta Field Cleanup lib/llm/src/preprocessor.rs	EOF flush path explicitly clears choice.delta.refusal and choice.delta.reasoning_content along with other reset fields before emitting reconstructed choice content.

---

🎯 1 (Trivial) | ⏱️ ~3 minutes

Tip

💬 Introducing Slack Agent: The best way for teams to turn conversations into code.

Slack Agent is built on CodeRabbit's deep understanding of your code, so your team can collaborate across the entire SDLC without losing context.

• Generate code and open pull requests
• Plan features and break down work
• Investigate incidents and troubleshoot customer tickets together
• Automate recurring tasks and respond to alerts with triggers
• Summarize progress and report instantly

Built for teams:

• Shared memory across your entire org—no repeating context
• Per-thread sandboxes to safely plan and execute work
• Governance built-in—scoped access, auditability, and budget controls

One agent for your entire SDLC. Right inside Slack.

👉 Get started

---

Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out.

❤️ Share

• X
• Mastodon
• Reddit
• LinkedIn

_{Comment @coderabbitai help to get the list of available commands and usage tips.}