[Models] add fleet model fallback 2

[xiaoguoguo626807]

Motivation

新增 PaddleFleet 作为模型推理后端（--model-impl paddlefleet），通过将 PaddleFleet TransformerLayer 中的 core_attention 替换为 FastDeploy Attention 内核，实现在 PaddleFleet 模型结构上复用 FastDeploy 的 KV Cache 和高性能 Attention 计算。 #7732 的清晰版

Modifications

• config.py: 新增 paddlefleet 到 ModelImpl 类型定义
• engine/args_utils.py: 支持 --model-impl paddlefleet CLI 参数，并补充校验逻辑
• model_executor/models/paddleformers/base_fleet.py: 新增 PaddleFleetModelBase 基类、FastDeployAttention 层及 patch_paddlefleet_core_attention 替换函数
• model_executor/models/paddleformers/__init__.py: 注册 PaddleFleetForCausalLM 模型类
• test_fallback_fleet_model.py` 需要独立的 PaddleFormers 和 PaddleFleet 依赖,使用 pytest conftest.py 钩子机制，在测试运行时动态安装依赖，避免污染全局环境

Usage or Command

python -m fastdeploy.entrypoints.openai.api_server \
    --model /path/to/model \
    --model-impl paddlefleet

Accuracy Tests

N/A（本 PR 新增 PaddleFleet 推理后端，尚未提供与参考实现的 logits 对齐数据）

Checklist

• Add at least a tag in the PR title.
  • Tag list: [[FDConfig],[APIServer],[Engine], [Scheduler], [PD Disaggregation], [Executor], [Graph Optimization], [Speculative Decoding], [RL], [Models], [Quantization], [Loader], [OP], [KVCache], [DataProcessor], [BugFix], [Docs], [CI], [Optimization], [Feature], [Benchmark], [Others], [XPU], [HPU], [GCU], [DCU], [Iluvatar], [Metax]]
  • You can add new tags based on the PR content, but the semantics must be clear.
• Format your code, run pre-commit before commit.
• Add unit tests. Please write the reason in this PR if no unit tests.
• Provide accuracy results.
• If the current PR is submitting to the release branch, make sure the PR has been submitted to the develop branch, then cherry-pick it to the release branch with the [Cherry-Pick] PR tag.

[paddle-bot]

Thanks for your contribution!

[PaddlePaddle-bot]

🤖 Paddle-CI-Agent | ci_status_monitor | 2026-06-02 20:14:56

CI报告基于以下代码生成（30分钟更新一次）:
PR commit: 80e85f8 | Merge base: 4474188 (branch: develop)

---

1 Required任务 : 8/10 通过

总执行（rerun次数）	总任务	✅ 通过	❌ 失败	⏳ 运行中	⏸️ 等待中	跳过
41(0)	41	37	2	1	1	0

任务	错误类型	置信度	日志
Approval	需要 Approval	高	Job

2 失败详情

🔴 Approval — 需要 Approval（置信度: 高）

该 Job 需要人工 Approval，完成审批后 CI 才会继续执行。

建议操作：请通过人工审批。

[codecov-commenter]

Codecov Report

❌ Patch coverage is 96.91358% with 10 lines in your changes missing coverage. Please review.
⚠️ Please upload report for BASE (develop@3fe8f7c). Learn more about missing BASE report.

Files with missing lines	Patch %	Lines
.../model_executor/models/paddleformers/base_fleet.py	96.68%	1 Missing and 9 partials ⚠️

Additional details and impacted files

@@            Coverage Diff             @@
##             develop    #7964   +/-   ##
==========================================
  Coverage           ?   67.56%           
==========================================
  Files              ?      468           
  Lines              ?    65903           
  Branches           ?    10169           
==========================================
  Hits               ?    44525           
  Misses             ?    18540           
  Partials           ?     2838           

Flag	Coverage Δ
GPU	77.67% <96.91%> (?)
XPU	7.02% <0.30%> (?)

Flags with carried forward coverage won't be shown. Click here to find out more.

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.

[PaddlePaddle-bot]

🤖 Paddle-CI-Agent | pr_review | 2026-06-03 13:56:29

📋 Review 摘要

PR 概述：新增 paddlefleet 作为模型推理后端，通过将 PaddleFleet TransformerLayer 中的 core_attention 替换为 FastDeploy Attention 内核，实现对 KV Cache 和高性能 Attention 的复用
变更范围：model_executor/models/paddleformers/、model_executor/graph_optimization/、engine/args_utils.py、config.py、tests/model_executor/fallback/
影响面 Tag：[Models] [FDConfig] [Graph Optimization] [CI] [Engine]

问题

级别	文件	概述
🟡 建议	fastdeploy/model_executor/models/paddleformers/base_fleet.py	assert 用于运行时校验，Python -O 下将静默失效

历史 Findings 修复情况

Finding	问题	状态
F1	args_utils.py help 字符串拼接缺少空格分隔	⚠️ 仍存在
F2	graph_opt_backend 不支持 *args，CUDAGraph 开启时可能 TypeError	🔄 部分修复
F3	multi_latent_attention 直接访问可能 AttributeError	✅ 已修复
F4	conftest.py 日志版本号与实际安装版本不一致	⚠️ 仍存在
F5	graph_opt_backend(*args, **kwargs) 传递 *args 时 GraphOptBackend.__call__ 仅接受 **kwargs，CUDAGraph 开启时仍会 TypeError	⚠️ 仍存在

📝 PR 规范检查

符合规范。

总体评价

F3（MLA AttributeError）已修复，整体实现逻辑清晰。历史 F1/F4 仍未修复，F5 核心问题（GraphOptBackend.__call__ 不接受 *args）仍存在，CUDAGraph 路径下启用 paddlefleet 会报 TypeError；建议跟进或在 PR 中明确说明 paddlefleet 暂不支持 CUDAGraph 模式。新增代码中 assert forward_meta is not None 需替换为显式 raise。

[gongshaotian]

LGTM for GraphOptBackend

[xiaoguoguo626807]

/re-run all-failed