Remove product-specific evaluations

[rgarcia]

summary

• remove application-specific evaluation tasks from the environment backend package
• remove the unrelated reporting integration extra and documentation
• keep CI focused on the backend package and its tests

validation

• uv run ruff format --check .
• uv run ruff check .
• uv run ty check
• uv run pytest (13 passed)
• uv build

---

Note

Low Risk
Documentation and dev/CI dependency cleanup only; no changes to the Hypeman environment backend runtime behavior.

Overview
Removes application-specific Harbor eval tasks under evals/kernel-mcp-server (browser-region and client-capabilities), including Docker environments, instructions, and Bun verifier tests.

Drops the optional braintrust extra from pyproject.toml, updates uv.lock, and simplifies CI from uv sync --locked --dev --extra braintrust to uv sync --locked --dev.

Deletes the README Braintrust reporting section and example harbor run commands that used those pinned kernel-mcp-server tasks.

^{Reviewed by Cursor Bugbot for commit 216f55f. Bugbot is set up for automated code reviews on this repo. Configure here.}