chore(automation): trim docstrings to intent only

Cut the docstrings and Field(description=...) text across the entire
automations/ tree down to single-line intent statements, matching the
multi_agent_chat conciseness style:

- Module docstrings: one line stating what the file is.
- Class docstrings: deleted when the class name + module docstring
  already cover intent; kept only where they add a constraint or
  rationale not visible in the signature.
- Pydantic Field descriptions: short noun phrases / clauses, not
  full sentences. Reasoning that belonged in the design plan moved
  out of the code.
- Enum values: per-value docstrings replaced with terse inline
  comments where the meaning isn't obvious from the name.

Behaviour is unchanged. The same 33 files, same public surface, same
imports — verified by re-running the 10-point registry smoke test and
the 8-point schema round-trip / constraint suite from commits 9 and
10.

LOC: 1180 → 691 (-42%).

surfsense_backend/app/automations/registries/capabilities/__init__.py

@@ -1,4 +1,4 @@
-"""Capability registry: ``types.py`` (dataclass), ``store.py`` (dict + register fn)."""
+"""Capability registry."""
 
 from __future__ import annotations
 

surfsense_backend/app/automations/registries/capabilities/store.py

@@ -1,4 +1,4 @@
-"""Capability registry: in-memory dict + ``register_capability`` API."""
+"""In-memory capability registry. Populated once at process startup."""
 
 from __future__ import annotations
 
@@ -8,33 +8,16 @@ _REGISTRY: dict[str, Capability] = {}
 
 
 def register_capability(capability: Capability) -> None:
-    """Add a capability to the in-memory registry.
-
-    Raises ``ValueError`` on duplicate ``id`` — registration is
-    idempotent only at the module level (a module's
-    ``register_capability`` call runs once per process), so a
-    duplicate is always a bug.
-    """
-
+    """Register a capability. Raises on duplicate id."""
     if capability.id in _REGISTRY:
-        raise ValueError(
-            f"Capability already registered: {capability.id!r}"
-        )
+        raise ValueError(f"Capability already registered: {capability.id!r}")
     _REGISTRY[capability.id] = capability
 
 
 def get_capability(capability_id: str) -> Capability | None:
-    """Look up one capability by id. Returns ``None`` on miss."""
-
     return _REGISTRY.get(capability_id)
 
 
 def all_capabilities() -> dict[str, Capability]:
-    """Snapshot of the registry as a defensive copy.
-
-    Returned dict is safe to iterate while other code calls
-    ``register_capability`` (which v1 never does post-startup, but
-    the contract holds anyway).
-    """
-
+    """Defensive snapshot of the registry."""
     return dict(_REGISTRY)

surfsense_backend/app/automations/registries/capabilities/types.py

@@ -1,4 +1,4 @@
-"""``Capability`` dataclass — the v1-minimum five-field shape."""
+"""``Capability`` dataclass and handler signature. Locked at five fields for v1."""
 
 from __future__ import annotations
 
@@ -7,32 +7,10 @@ from dataclasses import dataclass
 from typing import Any
 
 CapabilityHandler = Callable[[dict[str, Any]], Awaitable[Any]]
-"""The signature every capability handler must satisfy.
-
-The handler is a closure that already holds whatever runtime context
-it needs (DB session, search-space scope, logger, etc.). The
-registry only passes through the caller's input dict — the same dict
-that was validated against ``input_schema``.
-"""
 
 
 @dataclass(frozen=True, slots=True)
 class Capability:
-    """The unit of "what SurfSense can do," consumed by every layer.
-
-    v1 keeps the dataclass to exactly five fields. Earlier drafts
-    considered ``name``, ``required_credentials``, ``side_effects``,
-    ``expected_duration_seconds``, and ``cost_estimate``; every one
-    of those has been removed until a concrete consumer feature
-    requires it (see ``automation-design-plan.md`` §3, decision v1).
-
-    The handler is a ready-to-call function. It does not receive a
-    context argument — context is bound at registration time by the
-    factory that builds the closure (so a capability returned to an
-    agent's tool list looks identical to one returned to an
-    automation's action runtime).
-    """
-
     id: str
     description: str
     input_schema: dict[str, Any]