runtime: resume the learned stance in think, instead of discarding it

engram_think_json built a NEUTRAL stance on every call — cog_stance_init
with a NULL id, all axis_gain 1.0, bias_dir NULL, reliability 0.5 — and
never loaded the stance the correspondence-beat had been persisting.

That mattered because the faculty enters engram_think ONLY through the
stance: axis_gain[k] warps the per-axis extents and bias_dir seeds the
steering direction. cog_stance_init stores the faculty NAME and nothing
reads it. So with a neutral stance, reason/abduce/induce/plan/analogize
were byte-identical output under different labels, and confidence was
pinned to 0.5 because GeoGradient.confidence IS stance->reliability.

The machinery already existed and only this call site ignored it.
engram_correspondence_beat_json resumes via cog_stance_from_node and
persists via cog_stance_to_node under "stance-<faculty>-<hub>". Every
beat's calibration was written and then thrown away on the next read.
Same defect as the NULL anchor fixed in #142, one line below: a neutral
argument collapsing a capability to a constant.

Resume the same id the beat writes, so learning compounds across beats and
cold boot. Fall back to neutral only when no stance exists — a genuine
uninformed prior rather than a discarded informed one.

Also emit stance_resumed, so confidence 0.5 from a learned-but-unreliable
stance is distinguishable from confidence 0.5 from "no stance exists".
That reporting gap is what let the neutral stance hide.

Verified against a clone of the production store (13,627 nodes):

  before beat, no stance     stance_resumed=false  confidence=0.5
  beat on a NON-keystone     brier 0.00458568 -> 0.00329654
                             reduction 28.11%, n_trials 6000,
                             reliability 0.930726, stance_written=true
  after beat                 stance_resumed=true   confidence=0.930726

Confidence now equals the learned reliability instead of the uninformed
prior. The keystone self-anchor correctly stays at 0.5 — calibration is
deliberately refused on protected identity regions, and that refusal is
now visible as resumed=true with confidence unchanged, rather than being
indistinguishable from the bug.

STILL OPEN: with no learned bias_dir the faculties remain identical in
direction. What distinguishes abduce from induce geometrically is a
design decision about how Neuron thinks, not a plumbing defect, and is
deliberately left to Will.

lang/runtime/el_runtime.c

@@ -13937,7 +13937,40 @@ static int eg_cog_is_keystone_seeds(const char* csv) {
 el_val_t engram_think_json(el_val_t seeds, el_val_t faculty) {
     GeoDescriptor* g = eg_geo_build_desc(EL_CSTR(seeds));
     if (!g) return eg_geo_err("geometry unavailable");
-    CogStance st; cog_stance_init(&st, NULL, EL_CSTR(faculty), g->hub_id, NULL, g);
+    /* RESUME THE LEARNED STANCE (2026-08-16 self-review). This built a NEUTRAL
+     * stance every call — all axis_gain 1.0, bias_dir NULL, reliability 0.5 —
+     * and never loaded the one the correspondence-beat had been persisting.
+     *
+     * That mattered because the faculty enters engram_think ONLY through the
+     * stance: `gain = stance->axis_gain[k]` warps the per-axis extents, and
+     * `stance->bias_dir` seeds the steering direction. cog_stance_init stores
+     * the faculty NAME but nothing reads it. So with a neutral stance,
+     * reason / abduce / induce / plan / analogize are the same function with
+     * different labels — measured, byte-identical output across all five —
+     * and `confidence` is pinned to the 0.5 uninformed prior, because
+     * GeoGradient.confidence is just stance->reliability.
+     *
+     * The machinery already existed and only this call site ignored it:
+     * engram_correspondence_beat_json resumes via cog_stance_from_node and
+     * persists via cog_stance_to_node under the id "stance-<faculty>-<hub>".
+     * Every beat's calibration was being written and then thrown away on the
+     * next read. Same defect as the NULL anchor directly above: a neutral
+     * argument collapsing a capability to a constant.
+     *
+     * Resume the same id the beat writes, so learning compounds across beats
+     * and cold boot. Fall back to neutral only when no stance exists yet —
+     * which is a genuine uninformed prior, not a discarded informed one. */
+    char sid[256];
+    snprintf(sid, sizeof sid, "stance-%s-%s",
+             EL_CSTR(faculty) ? EL_CSTR(faculty) : "reason",
+             g->hub_id ? g->hub_id : "region");
+    CogStance st; StoreNode prev; int resumed = 0;
+    if (g_engram_store && store_get_node(g_engram_store, sid, &prev) == 1) {
+        if (cog_stance_from_node(&prev, &st) == 0) resumed = 1;
+        store_node_free(&prev);
+    }
+    if (!resumed) cog_stance_init(&st, sid, EL_CSTR(faculty), g->hub_id, NULL, g);
+    else { free(st.id); st.id = strdup(sid); }
     GeoGradient grad;
 
     /* ANCHOR THE READ (2026-08-16 self-review). This passed NULL, and NULL is
@@ -13999,8 +14032,13 @@ el_val_t engram_think_json(el_val_t seeds, el_val_t faculty) {
     if (engram_think(g, anchor, &st, &grad) != 0) { free(anchor); cog_stance_free(&st); engram_geo_free(g); return eg_geo_err("think failed"); }
     free(anchor);
     JsonBuf b; jb_init(&b); char t[256];
-    snprintf(t, sizeof t, "{\"faculty\":\"%s\",\"n_support\":%d,\"magnitude\":%.6g,\"spread\":%.6g,\"confidence\":%.6g,\"dim\":%d",
-             EL_CSTR(faculty), grad.n_support, grad.magnitude, grad.spread, grad.confidence, grad.dim);
+    /* stance_resumed distinguishes an INFORMED read from an uninformed one.
+     * Without it, confidence 0.5 from a learned-but-unreliable stance and
+     * confidence 0.5 from "no stance exists" are indistinguishable — the same
+     * reporting gap that let the NULL anchor and the neutral stance hide. */
+    snprintf(t, sizeof t, "{\"faculty\":\"%s\",\"n_support\":%d,\"magnitude\":%.6g,\"spread\":%.6g,\"confidence\":%.6g,\"stance_resumed\":%s,\"dim\":%d",
+             EL_CSTR(faculty), grad.n_support, grad.magnitude, grad.spread, grad.confidence,
+             resumed ? "true" : "false", grad.dim);
     jb_puts(&b, t);
     int emit = grad.dim < 8 ? grad.dim : 8;
     jb_puts(&b, ",\"direction\":"); eg_geo_emit_vec(&b, grad.direction, emit);