docs(core,hosts): crate-level Stability section; fix comments that contradict the code

- tines-core's crate docs carry the spec §8 compatibility policy, as the
  README does.
- tines-bt: the module and BluerRadio::new docs claimed an optional or
  empty allowlist. The allowlist is exactly the configured peer (F3).
- tines: the poll loop called Store::flush "infallible". It returns a
  Result, and a failure is logged and retried on the next schedule.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

tines-bt/src/transport.rs

@@ -32,8 +32,9 @@
 //!
 //! Up to [`MAX_PEERS`] connections may be live at once, one per bound
 //! [`NodeId`]. A second connection to the same [`NodeId`] replaces the first.
-//! The configured `peer` becomes an optional allowlist; empty means "any tines
-//! advertiser".
+//! Today the allowlist is exactly the configured `peer`, and frames and events
+//! from any other peer are dropped; an optional allowlist (empty meaning "any
+//! tines advertiser") is still to come (THREAT_MODEL.md F3).
 
 use std::collections::{BTreeMap, BTreeSet, HashMap};
 use std::sync::{
@@ -266,9 +267,9 @@ impl BluerRadio {
     /// Start the `bluer`/BlueZ radio for a connection between `me` (this
     /// device) and `peer` (the peer's advertised [`NodeId`]).
     ///
-    /// `peer` is this node's allowlist: only that NodeId's connections are
-    /// accepted. With an empty allowlist the transport accepts any tines
-    /// advertiser.
+    /// `peer` is this node's whole allowlist: only that NodeId's connections
+    /// are accepted. (The radio loop treats an empty allowlist as "any tines
+    /// advertiser", but `new` never builds one; THREAT_MODEL.md F3.)
     pub fn new(me: NodeId, peer: NodeId) -> Self {
         let (event_tx, event_rx) = async_mpsc::unbounded_channel::<RadioEvent>();
         let (send_tx, send_rx) = async_mpsc::unbounded_channel::<(NodeId, Vec<u8>)>();

tines-core/src/lib.rs

@@ -1,4 +1,30 @@
 //! tines-core: content-defined Observation identity and shared-memory core.
+//!
+//! # Stability
+//!
+//! tines-core 1.0 guarantees, within the 1.x line:
+//!
+//! - **Wire:** every 1.x release speaks wire v1 ([`consts::WIRE_VERSION`]).
+//!   Only new optional frame variants may be added, plus new [`SigAlgo`] ids
+//!   and signature versions within the 64-byte [`consts::MAX_SIG`] budget. A
+//!   wire-v2 change, including any post-quantum signature or a
+//!   [`consts::MAX_FRAME`] above 1500, ships as tines-core 2.0.
+//! - **Identity:** [`Observation::id`] never changes within a major version,
+//!   and should not change across majors unless the hash itself is broken.
+//! - **API:** standard semver; MSRV bumps are minor. Public types are
+//!   `#[non_exhaustive]` where they may grow. A method added to [`Store`],
+//!   [`Transport`] or [`BeliefPolicy`] after 1.0 must have a default whose
+//!   behaviour is safe: it may lose performance or telemetry, never data or
+//!   custody. Wall-clock folds live in the opt-in [`WallClockStore`] trait,
+//!   not in [`Store`]. The `#[doc(hidden)]` `conformance` module is not part
+//!   of the API.
+//! - **Local behaviour** (capacities, timings, scheduling order, which
+//!   candidate goes first) can change in a minor release as long as the
+//!   protocol properties hold: custody always keeps at least one holder,
+//!   receipts never shorten world state, and every frame fits
+//!   [`Transport::mtu`]. That's why the local constants are crate-private.
+//! - **Signatures:** 1.0 carries them ([`Observation::sig`]); 1.1 signs and
+//!   verifies them, without breaking the wire or the API.
 #![no_std]
 #![deny(missing_docs)]
 

tines/src/ipc.rs

@@ -266,9 +266,9 @@ where
         }
     }
 
-    // On a schedule: trim stale observations and flush the store. Flush is
-    // `LogStore::flush`, reached through `Node::flush(now)`, which delegates to
-    // `Store::flush` — the infallible durability boundary.
+    // On a schedule: trim stale observations and flush the store. `Node::flush`
+    // delegates to `LogStore`'s `Store::flush` impl, the durability point; a
+    // failure is logged and the next scheduled flush retries.
     if host.rounds >= POLL_CAP {
         host.rounds = 0;
         node.trim(&TrimPolicy::default(), host.clock.now());