fix(rocm): apply RoPE for embedding models without KV cache

[siluzhou]

When kv_cache is None (embedding models), AiterPrefillImplAsm and AiterPrefillImplNonAsm skipped RoPE and went straight to _forward_varlen. This caused severe output divergence for models that require positional encoding (e.g. tbstars with rope_theta=10000).

Also update _forward_varlen to handle the (q, k_padded, v_padded) tuple returned by the RoPE C++ op via unpad_kv_vectorized.

[LLLLKKKK]

AI Code Review - PR #973

Status: LGTM

Summary: P0/0 · P1/0 · P2/1 · P3/0

lgtm ready to ci

Non-blocking Suggestions

P2

• 缺少 no-KV RoPE 路径的回归测试 @ rtp_llm/models_py/modules/factory/attention/rocm_impl/aiter.py:916
  • 建议：补一个 ROCm 单测或 smoke，构造 need_rope_kv_cache=True、kv_cache=None 的 AiterPrefillImplAsm/NonAsm 路径，并校验输出与带 RoPE 的参考实现一致。

Checklist Violations (2 fail / 97 total)

General Principles Checklist

• [6.1] Tests — 新逻辑有聚焦单测 + 相关集成/smoke 测试 → issue 缺少 no-KV RoPE 路径的回归测试
  新增 no-KV RoPE wrapper 路径和 tuple varlen 输入处理，但 diff 未新增覆盖 kv_cache=None + need_rope_kv_cache=True 的测试。

RTP-LLM Checklist

• [H] 测试与 CI — 测试覆盖充分：大重构等价覆盖，新功能端到端测试 → issue 缺少 no-KV RoPE 路径的回归测试
  新增 no-KV RoPE 修复没有随 diff 增加对应 ROCm 单测或 smoke 覆盖。

Strengths

• 修复点同时覆盖 ASM 与 NonAsm 的 prefill no-KV 分支，避免两个 ROCm 后端行为分叉。
• _forward_varlen 复用现有 unpad_kv_vectorized，没有重新引入逐 batch Python unpad 循环。

[LLLLKKKK]

AI Code Review - PR #973

Status: LGTM

Summary: P0/0 · P1/0 · P2/0 · P3/0

lgtm ready to ci

Checklist ✅ (81 items passed)

Strengths

• ASM 与 NonASM 的 no-KV prefill 分支现在都会先执行 RoPE，再进入 FMHA，修复范围一致。
• 新增 wrapper 回归测试覆盖了 need_rope_kv_cache=True 且 kv_cache=None 时 RoPE 输出被传给 FMHA 的调用顺序。
• _forward_varlen 复用已有 unpad_kv_vectorized，没有重新引入逐 batch Python unpad 循环。

[wht21]

internal source has been updated, please review the changes!

[wht21]

internal source has been updated, please review the changes!