feat(profiler/behave_shell): emit motor.input_modality

BEHAVE-EXTRACTOR.md Phase A Step 2. The first primitive — picked
first because it has the highest discriminative value (HUMAN vs
everyone) and the simplest implementation (paste-event ratio over
total inputs).

* _features/motor.py:input_modality(ctx) emits one Observation
  per session in {typed, pasted, mixed} with confidence 0.75 / 0.70.
* _features/_emit.py centralises the make_observation helper so
  every feature module gets the same Window/source/evidence_ref
  boilerplate without copy-paste.
* Thresholds inherited from the prototype's calibration history
  (MODALITY_PASTED_MIN=0.40, MODALITY_TYPED_MAX=0.05).
* Zero-input session skips emission — registry doesn't admit
  "unknown" here.

Tests: pure-typed → typed, pure-pasted → pasted, mixed → mixed,
output-only session → no observation, full envelope round-trip.

decnet/profiler/behave_shell/_features/_emit.py

@@ -0,0 +1,32 @@
+"""Helper for building registry-valid :class:`Observation` records.
+
+Every feature module would otherwise repeat the same Window /
+source / evidence_ref boilerplate. This helper centralises it and is
+the one place to reach when emission semantics change (e.g. when we
+start parametrising windows on a per-primitive basis).
+"""
+from __future__ import annotations
+
+from typing import Any
+
+from decnet_behave_core.spec.envelope import Observation, Window
+
+from decnet.profiler.behave_shell._ctx import SessionContext
+
+
+def make_observation(
+    ctx: SessionContext,
+    *,
+    primitive: str,
+    value: Any,
+    confidence: float,
+) -> Observation:
+    """Build one :class:`Observation` for the whole-session window."""
+    return Observation(
+        primitive=primitive,
+        value=value,
+        confidence=confidence,
+        window=Window(start_ts=ctx.t_start, end_ts=ctx.t_end),
+        source=ctx.source,
+        evidence_ref=ctx.evidence_ref,
+    )