[megatron] fix: ValueError when unpacking preprocess_thd_engine result in router replay (verl-project#5891)

[megatron] fix: ValueError when unpacking preprocess_thd_engine result
in router replay

### What does this PR do?

Fixes `ValueError: too many values to unpack (expected 2) in
merge_router_topk_indices` when using router replay (R2) with non-legacy
workers.

`preprocess_thd_engine` returns a 3-tuple `(tensor, PackedSeqParams,
Optional[Tensor])` but the call site only unpacked 2 values.

This fix is necessary to run R2 router replay with non-legacy engine
workers, (in particular in fully async mode).

### Checklist Before Starting

- [x] Search for similar PRs. Paste at least one query link here:
[preprocess_thd_engine](https://github.com/verl-project/verl/pulls?q=is%3Apr+preprocess_thd_engine)
[R2](https://github.com/verl-project/verl/pulls?q=is%3Apr+R2) [router
replay](https://github.com/verl-project/verl/pulls?q=is%3Apr+router+replay)
- [x] Format the PR title as `[{modules}] {type}: {description}` (This
will be checked by the CI)
- `{modules}` include `fsdp`, `megatron`, `veomni`, `sglang`, `vllm`,
`rollout`, `trainer`, `ci`, `training_utils`, `recipe`, `hardware`,
`deployment`, `ray`, `worker`, `single_controller`, `misc`, `perf`,
`model`, `algo`, `env`, `tool`, `ckpt`, `doc`, `data`, `cfg`, `reward`,
`fully_async`, `one_step_off`
- If this PR involves multiple modules, separate them with `,` like
`[megatron, fsdp, doc]`
  - `{type}` is in `feat`, `fix`, `refactor`, `chore`, `test`
- If this PR breaks any API (CLI arguments, config, function signature,
etc.), add `[BREAKING]` to the beginning of the title.
  - Example: `[BREAKING][fsdp, megatron] feat: dynamic batching`

### Test

This is a one-line fix matching the function's return type annotation.
No existing tests cover this code path (router replay with nested
tensors). The fix was validated by inspecting the
`preprocess_thd_engine` signature at `verl/models/mcore/util.py` which
returns `tuple[torch.Tensor, PackedSeqParams, Optional[torch.Tensor]]`.

### API and Usage Example

No API changes.

### Design & Code Changes

No design changes, just a one-line fix.

### Checklist Before Submitting

> [!IMPORTANT]
> Please check all the following items before requesting a review,
otherwise the reviewer might deprioritize this PR for review.

- [x] Read the [Contribute
Guide](https://github.com/volcengine/verl/blob/main/CONTRIBUTING.md).
- [x] Apply [pre-commit
checks](https://github.com/volcengine/verl/blob/main/CONTRIBUTING.md#code-linting-and-formatting):
`pre-commit install && pre-commit run --all-files --show-diff-on-failure
--color=always`
- [x] Add / Update [the
documentation](https://github.com/volcengine/verl/tree/main/docs). (not
applicable)
- [x] Add unit or end-to-end test(s) to [the CI
workflow](https://github.com/volcengine/verl/tree/main/.github/workflows)
to cover all the code. If not feasible, explain why: just a bug fix
- [ ] Once your PR is ready for CI, send a message in [the `ci-request`
channel](https://verl-project.slack.com/archives/C091TCESWB1) in [the
`verl` Slack
workspace](https://join.slack.com/t/verl-project/shared_invite/zt-3855yhg8g-CTkqXu~hKojPCmo7k_yXTQ).
(If not accessible, please try [the Feishu group
(飞书群)](https://applink.larkoffice.com/client/chat/chatter/add_by_link?link_token=772jd4f1-cd91-441e-a820-498c6614126a).)
- [x] If your PR is related to the `recipe` submodule, please also
update the reference to the submodule commit via `git submodule update
--remote` or `cd recipe && git pull origin main`. (not applicable)

Author: guillemgt

verl/utils/megatron/router_replay_utils.py

@@ -253,7 +253,7 @@ def merge_router_topk_indices(attention_mask, input_ids, mini_layer_topk_idx_lis
 
         if input_ids.is_nested:
             batch_size = input_ids.shape[0]
-            _, packed_seq_params = preprocess_thd_engine(input_ids, pre_process=True)
+            _, packed_seq_params, _ = preprocess_thd_engine(input_ids, pre_process=True)
             layers_topk_idx = postprocess_thd_engine(
                 layers_topk_idx, packed_seq_params, input_ids, batch_size, post_process=True
             )