Add configurable trusted-MCP presets

[set2374]

Right now trusted MCP tool detection is regex-driven and intentionally conservative. It would be useful to make this easier to tune for common environments.\n\nSuggested scope:\n- define a few preset bundles in docs or config examples\n- separate 'authoritative data' MCPs from 'workflow convenience' MCPs\n- document the tradeoff between broader trust patterns and false verification signals\n\nThis issue is a good fit for contributors who care about practical deployment hygiene.

[snakefood3232]

The regex-driven trusted MCP detection is indeed too conservative for common environments. I'd handle this by implementing configurable preset bundles that separate authoritative data sources from workflow tools, with clear documentation on trust trade-offs. Can have a PR up in 24 hours—I've built similar config systems for Python automation pipelines.