dream_observation: drop the 'go quiet' rule from select_mode

The earlier behavior never went quiet — it dreamed every night, even when
that meant repeating itself. The 'return None on null delta' rule was a
synthesis-doc invention (the dreamer-design-spec.md I treated as
authoritative is itself LLM-generated) that didn't match the actual
desired UX. Aaron called this out.

The repetition problem the quiet rule was claimed to solve is already
addressed in the retrieve layer:
- LLM-generated queries from the observation signal vary nightly
- MMR diversity prevents within-night cluster lock-in
- NREM bias toward under-processed chunks (low consolidation_count)
  ensures fresh material gets selected over recently-replayed material

So select_mode now always returns a mode. NREM is the default. Staleness
still routes to Late REM at 3+ days for cross-domain variety. Journal
entries still route to Early REM.

scripts/dream_observation.py

@@ -186,50 +186,38 @@ def observe_corpus():
 # ─── Stage 2: select_mode ───────────────────────────────────────────────────
 
 def select_mode(signal, task=None, explicit_mode=None):
-    """Return one of {'nrem', 'early-rem', 'late-rem', 'lucid'} or None.
+    """Return one of {'nrem', 'early-rem', 'late-rem', 'lucid'}. Never None.
 
-    Selection logic from spec (lines 69–74):
-      - Explicit mode argument → use that mode
+    The dreamer fires every scheduled night. The earlier "go quiet on null
+    delta" rule was a synthesis-doc invention that didn't match the actual
+    desired UX — the original dreamer always dreamed, even if it repeated
+    itself. The cure for repetition lives in the retrieve layer
+    (LLM-generated queries from the observation signal, MMR diversity,
+    cursor bias toward under-processed chunks), not in skipping nights.
+
+    Routing logic:
+      - explicit_mode argument wins
       - task supplied → 'lucid' (question-anchored)
-      - Active journal entry → Early REM (emotional/personal register)
-      - Corpus unchanged ≥ STALENESS_TRIGGER_DAYS → Late REM (shake loose)
-      - New chunks above threshold → NREM
-      - Otherwise → None ("dreamer goes quiet rather than manufacturing novelty")
-
-    The None return is load-bearing. Per spec line 67, it's the canonical
-    answer to the repetition problem: when nothing has changed, do not
-    manufacture a dream just to fill the schedule slot.
+      - days_since_dream ≥ STALENESS_TRIGGER_DAYS → 'late-rem' (shake loose
+        via cross-domain pairs when nothing's been added in a while)
+      - new journal entry → 'early-rem' (emotional/personal register)
+      - default → 'nrem' (replay-and-consolidation; always has something to
+        do because the corpus always has under-processed chunks)
     """
     if explicit_mode:
         return explicit_mode
     if task:
         return "lucid"
 
-    new_chunks = signal["new_chunks"]
-    new_journal = signal["new_journal_entries"]
     days_since = signal["days_since_dream"]
+    new_journal = signal["new_journal_entries"]
 
-    # Spec line 72: corpus unchanged ≥3 days → Late REM ("shake things loose").
-    # This rule has to win against the "go quiet" rule below — the spec uses
-    # quiet as the default for 1-2 days of silence, then deliberately fires
-    # Late REM at 3+ days to break the stasis. Checking staleness first is
-    # what implements that intent. (Previous version checked "go quiet" first
-    # and never reached this branch.)
     if days_since >= STALENESS_TRIGGER_DAYS:
         return "late-rem"
 
-    # Spec line 71: journal entry → Early REM
-    # We treat "any new journal entry" as the trigger. Refining to detect
-    # emotional/personal register requires LLM sentiment tagging, which
-    # is left for a later iteration.
     if new_journal:
         return "early-rem"
 
-    # Spec line 67: nothing changed within the staleness window → quiet
-    if new_chunks < NEW_CHUNK_THRESHOLD and not new_journal:
-        return None
-
-    # Default: new chunks above threshold → NREM
     return "nrem"