feat: evaluate multiple targets

[christso]

Problem

Prompts behave differently depending on the provider. For example, VS Code uses runSubagent too, while Copilot CLI uses task agent.
Also, Copilot CLI has GitHub MCP built-in by default, while VS Code has githubRepo tool built-in.

Proposal

In the eval yaml, allow user to define multiple targets to ensure prompts work across multiple targets.