fix: align red-team-benchmark with actual adversarial_dojo CLI (#4138)

* fix: align red-team-benchmark with actual adversarial_dojo CLI

The workflow was using incorrect CLI flags and config schema:
- search-attacks takes a positional config arg, not --config
- Output dir flag is --out, not --output-dir
- No --agentshim CLI flag exists; AWF wraps the entire run
- TOML config must use [agents.victim] schema with tool_interface_file

Also fixes validation grep checks to match actual --help output
(--out, --victim-provider) and updates test assertions.

Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>

* fix: address red-team benchmark review feedback

---------

Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>
Co-authored-by: copilot-swe-agent[bot] <198982749+Copilot@users.noreply.github.com>

Author: 3 people