refactor(web): per-capability backend selection for search/extract split

Introduce the foundation for independently selecting web search and
extract backends — enabling future combinations like SearXNG for
search + Firecrawl for extract.

Architecture:
- tools/web_providers/base.py: WebSearchProvider and WebExtractProvider
  ABCs with normalized result contracts (mirrors CloudBrowserProvider)
- tools/web_tools.py: _get_search_backend() and _get_extract_backend()
  read per-capability config keys, fall through to shared web.backend
- hermes_cli/config.py: web.search_backend and web.extract_backend in
  DEFAULT_CONFIG (empty = inherit from web.backend)

Behavioral change:
- web_search_tool() now dispatches via _get_search_backend()
- web_extract_tool() now dispatches via _get_extract_backend()
- When per-capability keys are empty (default), behavior is identical
  to before — _get_search_backend() falls through to _get_backend()

This is purely structural — no new backends are added. SearXNG and
other search-only/extract-only providers can now be added as simple
drop-in modules in follow-up PRs.

12 new tests, 49 existing tests pass with zero regressions.

Ref: #19198

tools/web_tools.py

@@ -119,7 +119,7 @@ def _load_web_config() -> dict:
         return {}
 
 def _get_backend() -> str:
-    """Determine which web backend to use.
+    """Determine which web backend to use (shared fallback).
 
     Reads ``web.backend`` from config.yaml (set by ``hermes tools``).
     Falls back to whichever API key is present for users who configured
@@ -145,6 +145,44 @@ def _get_backend() -> str:
     return "firecrawl"  # default (backward compat)
 
 
+def _get_search_backend() -> str:
+    """Determine which backend to use for web_search specifically.
+
+    Selection priority:
+    1. ``web.search_backend`` (per-capability override)
+    2. ``web.backend`` (shared fallback — existing behavior)
+    3. Auto-detect from env vars
+
+    This enables using different providers for search vs extract
+    (e.g. SearXNG for search + Firecrawl for extract).
+    """
+    return _get_capability_backend("search")
+
+
+def _get_extract_backend() -> str:
+    """Determine which backend to use for web_extract specifically.
+
+    Selection priority:
+    1. ``web.extract_backend`` (per-capability override)
+    2. ``web.backend`` (shared fallback — existing behavior)
+    3. Auto-detect from env vars
+    """
+    return _get_capability_backend("extract")
+
+
+def _get_capability_backend(capability: str) -> str:
+    """Shared helper for per-capability backend selection.
+
+    Reads ``web.{capability}_backend`` from config; if set and available,
+    uses it. Otherwise falls through to the shared ``_get_backend()``.
+    """
+    cfg = _load_web_config()
+    specific = (cfg.get(f"{capability}_backend") or "").lower().strip()
+    if specific and _is_backend_available(specific):
+        return specific
+    return _get_backend()
+
+
 def _is_backend_available(backend: str) -> bool:
     """Return True when the selected backend is currently usable."""
     if backend == "exa":
@@ -1129,8 +1167,8 @@ def web_search_tool(query: str, limit: int = 5) -> str:
         if is_interrupted():
             return tool_error("Interrupted", success=False)
 
-        # Dispatch to the configured backend
-        backend = _get_backend()
+        # Dispatch to the configured search backend
+        backend = _get_search_backend()
         if backend == "parallel":
             response_data = _parallel_search(query, limit)
             debug_call_data["results_count"] = len(response_data.get("data", {}).get("web", []))
@@ -1286,7 +1324,7 @@ async def web_extract_tool(
         if not safe_urls:
             results = []
         else:
-            backend = _get_backend()
+            backend = _get_extract_backend()
 
             if backend == "parallel":
                 results = await _parallel_extract(safe_urls)