Clarify sandbox-first guidance for terminal tool unsandboxing

[dileepyavan]

Fixes #304596

This updates the run-in-terminal tool guidance so models are told to execute commands in sandbox mode first and only set requestUnsandboxedExecution=true when retrying after a sandboxed failure or when blocked domains are explicitly required.

It also updates the related unit tests to lock in the new prompt and schema wording.

Validation:

• Focused RunInTerminalTool sandbox invocation messaging tests
• Hygiene on the touched files

[Copilot]

Pull request overview

Updates the runInTerminal tool’s model-facing guidance and input schema wording to encourage sandbox-first execution and to tighten when requestUnsandboxedExecution should be used, with accompanying unit test assertions to lock the prompt/schema text.

Changes:

• Added/updated sandboxing guidance in the tool model description to discourage unsandboxing before a sandbox attempt.
• Refined the JSON schema descriptions for requestUnsandboxedExecution and requestUnsandboxedExecutionReason to require concrete justification (sandbox failure or blocked-domain need).
• Updated RunInTerminalTool unit tests to assert the new guidance and schema text.

Show a summary per file

File	Description
src/vs/workbench/contrib/terminalContrib/chatAgentTools/browser/tools/runInTerminalTool.ts	Updates sandbox guidance text and schema descriptions to drive sandbox-first behavior and require explicit unsandbox justification.
src/vs/workbench/contrib/terminalContrib/chatAgentTools/test/electron-browser/runInTerminalTool.test.ts	Adds assertions verifying the updated model description and schema descriptions for sandbox/unsandbox guidance.

Copilot's findings

• Files reviewed: 2/2 changed files
• Comments generated: 1