refactor(routes): move search domain into routes/search/ subpackage

[ydonghao]

Summary

Slice 2j of the route-domain reorganization (Refs #4082 / #4071). Moves the search domain (111 lines) into routes/search/ behind a sys.modules shim. Pure file reorganization, no behavior change.

One of three parallel low-risk slices (search, vault, webhook). Zero source-introspection landmines. Canonical module imports only from services/ and stdlib.

How to test

python -m compileall routes/search/ routes/search_routes.py app.py
python -m pytest tests/ -q

Target branch

• This PR targets dev, not main.

Linked Issue

Follows the Phase 0 architecture refactor (#4082 / #4071). One domain per PR.

Type of Change

• Refactor / cleanup

Checklist

• I searched open issues and open PRs — this is not a duplicate
• This PR targets dev
• My changes are limited to the scope described above — no unrelated refactors
• No runtime behavior changes (pure file move + import shim)
• Backward-compatible: old import paths keep working via shim

[ydonghao]

Friendly ping for review 👋

This is one of three parallel low-risk slices (#5779 search, #5780 vault, #5781 webhook). Only the search domain moves — a single-file, 111-line domain with zero internal routes/ coupling — into routes/search/ behind the same sys.modules shim as all previous slices.

Both points from your #4903 review are baked in (canonical doesn't depend on shim; shim regression test included).

Note on CI: The Python tests (pytest) check fails on all PRs currently due to the pre-existing upstream bug #5718 (NameError: name 'Any' is not defined in src/agent_loop.py), not from this PR. Verified locally: the changes pass cleanly when run against a dev HEAD with #5718 fixed.

@RaresKeY @alteixeira20 — would you mind taking a look when you have a moment?

[RaresKeY]

LGTM