Docs: note 1s tool coalescing window

CHANGELOG.md

@@ -5,7 +5,7 @@
 ### Highlights
 - **Thinking directives & state:** `/t|/think|/thinking <level>` (aliases off|minimal|low|medium|high|max/highest). Inline applies to that message; directive-only message pins the level for the session; `/think:off` clears. Resolution: inline > session override > `inbound.reply.thinkingDefault` > off. Pi/Tau get `--thinking <level>` (except off); other agents append cue words (`think` → `think hard` → `think harder` → `ultrathink`). Heartbeat probe uses `HEARTBEAT /think:high`.
 - **Verbose directives + session hints:** `/v|/verbose on|full|off` mirrors thinking: inline > session > config default. Directive-only replies with an acknowledgement; invalid levels return a hint. When enabled, tool results from JSON-emitting agents (Pi/Tau, etc.) are forwarded as metadata-only `[🛠️ <tool-name> <arg>]` messages (now streamed as they happen), and new sessions surface a `🧭 New session: <id>` hint.
-- **Verbose tool coalescing:** successive tool results of the same tool within ~250ms are batched into one `[🛠️ tool] arg1, arg2` message to reduce WhatsApp noise.
+- **Verbose tool coalescing:** successive tool results of the same tool within ~1s are batched into one `[🛠️ tool] arg1, arg2` message to reduce WhatsApp noise.
 - **Directive confirmations:** Directive-only messages now reply with an acknowledgement (`Thinking level set to high.` / `Thinking disabled.`) and reject unknown levels with a helpful hint (state is unchanged).
 - **Pi/Tau stability:** RPC replies buffered until the assistant turn finishes; parsers return consistent `texts[]`; web auto-replies keep a warm Tau RPC process to avoid cold starts.
 - **Claude prompt flow:** One-time `sessionIntro` with per-message `/think:high` bodyPrefix; system prompt always sent on first turn even with `sendSystemOnce`.