feat(web): create_automation HITL approval card in chat

Closes the create loop in chat: the agent describes user intent → the
drafter sub-LLM produces an AutomationCreate JSON → this card surfaces
a structured preview → approve persists; reject cancels. Edits flow
through chat refinement (re-call with a refined intent), not in-card,
so the card stays simple and the multi-turn checkpointer carries the
context.

Tool UI (components/tool-ui/automation/):
- create-automation.tsx — entry dispatcher + ApprovalCard chrome
  (pending/processing/complete/rejected via useHitlPhase) + SavedCard
  (links to the detail page) + InvalidCard (lists drafter validation
  issues) + ErrorCard (verbatim message). Rejection result is hidden
  because the approval card itself shows the rejected phase inline.
- automation-draft-preview.tsx — structured preview body: name +
  description + goal, triggers (humanised cron + tz + static-input
  keys), plan steps (step_id → action), and a collapsible raw JSON
  for power users.

Wiring:
- components/tool-ui/index.ts — re-export.
- features/chat-messages/timeline/tool-registry/registry.ts —
  register create_automation → CreateAutomationToolUI (dynamic import,
  same pattern as other connector tools).
- contracts/enums/toolIcons.tsx — Workflow icon + "Create automation"
  display name so fallback chrome (and timeline headers) are honest.

Shared util:
- lib/automations/describe-cron.ts — lifted from the route slice's
  lib/ folder since both the dashboard slice and the new approval card
  now render schedule descriptions. Slice imports updated; the now-
  empty slice lib/ folder is gone.

Backend prompt fragments:
- main_agent/system_prompt/.../create_automation/description.md and
  the tool's docstring no longer promise in-card edits. They make the
  refinement path explicit: if the user wants changes after seeing the
  draft, they reply in chat and the agent calls the tool again with a
  refined intent.

v1 deliberately excludes:
- In-card edit form / right-side edit panel — defer until we see real
  demand. The chat refinement loop covers the common case.
- approve_always / persistent allow rules — automations are a single
  artifact, not a repeated mutation, so the "trust this kind of call"
  affordance doesn't apply.

surfsense_backend/app/agents/multi_agent_chat/main_agent/system_prompt/prompts/tools/create_automation/description.md

@@ -1,7 +1,7 @@
 - `create_automation` — Draft and author a new automation. You describe the
   user's intent; a focused drafter inside the tool turns it into the full
-  automation JSON; the user reviews and edits it on an approval card; on
-  approval it's saved. All three phases happen in a single tool call.
+  automation JSON; the user sees a preview on an approval card and chooses
+  approve or reject. All three phases happen in a single tool call.
   - Call when the user wants SurfSense to do something on its own: anything
     recurring or scheduled ("every morning…", "each Monday…", "weekly
     recap…").
@@ -17,13 +17,16 @@
         explicitly ("the Notion parent page id was not specified") so the
         drafter leaves a placeholder.
   - Do NOT prompt the user to confirm before calling — the approval card
-    IS the confirmation. The user can edit any field on the card.
+    IS the confirmation. The card shows a structured preview plus the raw
+    JSON; it offers approve/reject only. If the user wants changes after
+    seeing the draft, they reply in chat and you call this tool again with
+    a refined `intent` — that's the edit path.
   - Returns:
     - `{status: "saved", automation_id, name}` — confirm briefly to the
       user ("Saved as automation #N — runs <when>."). Don't dump JSON back.
     - `{status: "rejected", message}` — the user declined on the card.
       Acknowledge once ("Understood, I didn't create it.") and stop. Do
-      NOT retry or pitch variants.
+      NOT retry or pitch variants without a fresh user request.
     - `{status: "invalid", issues, raw?}` — drafting/validation failed
       before the card was shown. Read the issues, refine your `intent`
       with the missing details, call again.

surfsense_backend/app/agents/multi_agent_chat/main_agent/tools/automation/create.py

@@ -66,9 +66,11 @@ def create_create_automation_tool(
         names, …) it needs.
 
         The tool drafts the full automation JSON internally, shows the user
-        an approval card for review, and persists on approval. Do NOT
-        prompt the user to confirm before calling — the card IS the
-        confirmation. The user can edit any field there.
+        a structured preview on an approval card, and persists on approval.
+        The card supports approve/reject only — if the user wants edits
+        after seeing the draft, they say so in chat and you call this tool
+        again with a refined intent. Do NOT prompt the user to confirm
+        before calling — the card IS the confirmation.
 
         Args:
             intent: Concrete restatement of the user's request. Include