Agent Debug Logs Cache Explorer gets confused between multiple concurrent AI conversations/requests

```
Version: 1.137.0 (Universal)
Commit: 645f29cc3176500b4b5762ba887cf2a7f0ffdf2c
Date: 2026-09-08T13:43:59-07:00
Electron: 42.10.0
ElectronBuildId: 15109253
Chromium: 148.0.7778.280
Node.js: 24.18.1
V8: 14.8.178.38-electron.0
@github/copilot: 1.0.84.70.gdb75d0d
@github/copilot-sdk: 1.0.13-preview.4
OS: Darwin arm64 25.5.0
```

## Description

When interacting with VSCode Copilot Chat with Agent Debug Logs enabled, there are multiple concurrent "conversations" with the selected AI model - the main conversation, any subagents, and utility conversations (in this example, for tool summarisation).

The Cache Explorer view of Agent Debug Logs makes no distinction between them and sees everything as "copilotLanguageModelWrapper" - and interprets subsequent requests in a different "conversation" as a cache miss / history rewritten / system prompt rewritten.

This makes it difficult to reliably pick up on actual changes that are busting the cache vs just noise from switching conversation contexts.

## Screenshots
Tool summarization request

Subagent initiation

## Steps to reproduce
1. Turn on Agent Debug Logs
2. Have a conversation in local Agent mode with tools enabled, potentially invoking subagents
3. Open Agent Debug Logs > Cache Explorer
4. Notice the issue with the [cache 0%] and diffs in the system prompt