[WilhelmZA]

Review remediation

Follow-up to merged PR #205. Refs #128; the issue remains open for reviewer acceptance.

• Default the device prompt to no when structured_output is absent.
• Preserve an existing device's explicit opt-in when the operator presses Enter.
• Describe structured.enable_for_bgp_route and structured.enable_for_traceroute separately from device opt-in.
• Verify both prompt choices and preservation of existing opt-in through the interactive CLI tests.

Validation

• Onboarding tests: 22 passed.
• Ruff on changed Python files: passed.
• git diff --check: passed.
• CI passed.
• Both live onboard check reports reviewed against configuration files and model defaults: all 4 structured-instance findings and 26 text-instance findings confirmed.
• lab-ioll2 intentionally uses text on both instances. Both device files explicitly set structured_output: false; lab/tests/test_lab.py also requires text for this device.
• Full live run: all 15 configured appliances and four VRF locations queried on both configurations, with fixture feeds enabled. No location skipped. Optional appliances ran one at a time.
• Existing lab matrix: 171/171 same-device/query comparisons match. Expected empty results remain CLI failures and API warnings.
• Junos reachable-target retest: 8/8 queries and 8/8 supplementary API comparisons pass on each configuration.
• IOSv exposes one genuine structured AS-path parser failure in both the CLI and API. Its other five built-ins pass; all six pass in text mode. The evidence preserves this failure.
• Complete acceptance summary, structured per-query summaries, text per-query summaries.
• Test-only core fixtures and the Huawei feed-cap change are removed. Original routes remain. Lab BGP session states are restored; optional appliances are stopped.

Keep #128 open until the reviewer accepts the live evidence.