fix: bound Lark card text fallback outbound waits

[jason-aelf]

Summary

• Bound Lark card text fallback outbound waits so card delivery does not wait indefinitely on edit/send failures.
• Moved shared Lark outbound/card/text edit contracts into the platform Lark abstractions package and wired the concrete tool provider implementation.
• Added regression coverage for AgentRun Lark card delivery, outbound dispatch, text edit fallback, and scheduled skill-runner reply sinks.

Test Plan

• bash tools/ci/test_stability_guards.sh
• dotnet test test/Aevatar.AI.ToolProviders.Lark.Tests/Aevatar.AI.ToolProviders.Lark.Tests.csproj --nologo
• dotnet test test/Aevatar.GAgents.ChannelRuntime.Tests/Aevatar.GAgents.ChannelRuntime.Tests.csproj --nologo

Notes

• PR contains commit c96caa6.
• No merge was performed.

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 83.77%. Comparing base (d1ef994) to head (23bc4e2).
⚠️ Report is 19 commits behind head on feature/integrate.

@@                  Coverage Diff                  @@
##           feature/integrate    #2497      +/-   ##
=====================================================
+ Coverage              83.74%   83.77%   +0.03%     
=====================================================
  Files                   1244     1244              
  Lines                  95469    95512      +43     
  Branches               12524    12530       +6     
=====================================================
+ Hits                   79952    80019      +67     
+ Misses                  9954     9930      -24     
  Partials                5563     5563              

Flag	Coverage Δ
ci	83.77% <ø> (+0.03%)	⬆️

Flags with carried forward coverage won't be shown. Click here to find out more.
see 6 files with indirect coverage changes

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.
• 📦 JS Bundle Analysis: Save yourself from yourself by tracking and limiting bundle sizes in JS merges.

[louis4li]

我看了一下，这个 PR 方向是在解决正确的问题，但当前实现明显偏重，建议先收敛后再合。

目标可以更简单地表达为：

1. Lark 收到 workflow 请求后，先快速回一条普通文本，例如“正在处理”；
2. workflow 最终结果出来后，优先 best-effort 编辑第一条文本；
3. 如果没有编辑权限、编辑失败或 edit port 不可用，就直接发送新的普通文本；
4. 长内容按普通文本分段发送；
5. CardKit 只能是可选的 rich renderer，不能成为首条可见回复或最终结果投递的依赖。

当前 PR 的主要问题是：

• 实现仍然围绕 LarkCardTextFallback，看起来是在 card 创建失败后才进入文本 fallback；但用户问题是 workflow 等很久没有反馈，所以首条普通文本应该在 workflow accepted / durable handoff 时就发出，而不是等 card 路径失败或超时。
• 改动面过大：新增/迁移 Lark platform abstractions、proxy response、error codes、edit port、docs 和大量测试。这里有些抽象可能有独立价值，但不应该和这个窄修复绑在一个 PR 里。
• 引入了过重的 fallback 状态机和持久状态：phase、segment count、last delivered text、error fields、fallback changed event、timeout/update/retry 等。这些更像 transport delivery telemetry，不应该成为 AgentRun 的核心 domain state。这个需求最多需要持久化第一条 text message id / correlation，以及最终文本是否完成投递。
• edit 不应该是硬依赖。正确语义是 edit 成功就编辑第一条，edit 不可用或失败就发新文本；不能因为缺少 card/edit 权限导致普通文本最终结果也失败。

建议改成最小实现：

Lark message received
  -> workflow/run accepted or durable handoff
  -> send ordinary text: 正在处理...
  -> store message id if available

workflow final text ready
  -> split by Lark plain-text limit
  -> try edit first processing message with first segment
  -> if edit unavailable/fails, send first segment as new text
  -> send remaining segments as ordinary text
  -> CardKit optional only

建议本 PR 收敛为上述普通文本主路径；Platform.Lark.Abstractions 迁移如果确实需要，建议拆成独立 PR，否则当前 review/回归面过大。