docs: clarify API server tool execution locality (salvage #19117)

[teknium1]

Closes #19117 via salvage. First increment for #18715.

Summary

• Clarifies that Hermes' OpenAI-compatible API server is a server-side Hermes agent runtime, not a pure LLM proxy (tools execute on the API-server host).
• Adds runtime metadata to /v1/capabilities so clients can detect the server_agent / tool_execution: server semantics.
• Doc update to open-webui.md reflecting same.

Validation

scripts/run_tests.sh tests/gateway/test_api_server.py → 138 passed.

Original author: @pingchesu.

[github-actions]

🔎 Lint report: salvage/pr-19117 vs origin/main

ruff

Total: 0 on HEAD, 0 on base (➖ 0)

🆕 New issues: none

✅ Fixed issues: none

Unchanged: 0 pre-existing issues carried over.

ty (type checker)

Total: 7498 on HEAD, 7498 on base (➖ 0)

🆕 New issues: none

✅ Fixed issues: none

Unchanged: 3938 pre-existing issues carried over.

Diagnostics are surfaced as warnings — this check never fails the build.