Orphan discs in regions + orphan exemption from depth anchoring

Two follow-ups on the semantic-cluster wave, both user-reported:

1. Region coloring: edge-only dilation (the anti-double-counting rule
   from the Louvain era) left 114 of 338 cluster members completely
   uncolored — semantic clusters contain members that share no
   intra-cluster edge with anyone (90 edge-orphans + isolated
   micro-topics), and 20 of 45 clusters painted nothing at all. Fix:
   a disc (radius = dilation, same as a round cap) for exactly the
   members no intra-cluster edge stroke reached. Every member is
   painted by its edges' caps OR one disc, never both — the
   no-double-counting rule survives. Verified live: all 74
   orphans-in-clusters sit on tinted region pixels.

2. Depth modes (ORPHAN RULE): a node with no edges has no dependency
   structure, so depth-0 stamping put 90 unconnected propositions on
   the vertical baseline / radial base ring — meaning they don't have.
   Orphans are now exempt from the depth force and get the regular
   view's unprojected forces instead (full center gravity + full
   cluster pull). Structured-node projections unchanged (x-only
   gravity in vertical, none in radial; tangential-only cluster pull
   in radial). Gravity in depth modes is a custom force with EXACTLY
   d3's forceX/forceY math (strength × alpha — the first draft
   forgot the alpha and shifted the settle). Verified live: radial
   orphans follow their topics instead of the base ring; vertical
   orphans spread 1006 world units (6/90 near the baseline band vs
   90/90 before).

components/graph-canvas.ts

@@ -1205,10 +1205,9 @@ export class GraphCanvasRenderer {
 		});
 
 		// The region IS the dilated skeleton: intra-cluster edges as
-		// round-cap strokes (Minkowski dilation — points become discs at
-		// line ends via the caps). NO per-node discs: a cluster is its
-		// connected points, and discs double-counted where nodes sit on
-		// edges (user-identified hot spots).
+		// round-cap strokes (Minkowski dilation). NO blanket per-node discs
+		// — they double-counted where nodes sit on edges (user-identified
+		// hot spots, when clusters were edge-based).
 		this.regionMeta = [];
 		data.clusters.forEach((cluster, ci) => {
 			const hue = clusterHue(cluster.id);
@@ -1217,6 +1216,8 @@ export class GraphCanvasRenderer {
 			let members = 0;
 			let sumX = 0;
 			let sumY = 0;
+			// Endpoints actually covered by a stroke's round caps.
+			const touched = new Set<string>();
 			for (const e of data.edges) {
 				if (clusterOf.get(e.from) !== ci) continue;
 				if (clusterOf.get(e.to) !== ci) continue;
@@ -1225,9 +1226,29 @@ export class GraphCanvasRenderer {
 				if (!a || !b) continue;
 				rctx.moveTo((a.x - minX) / cell, (a.y - minY) / cell);
 				rctx.lineTo((b.x - minX) / cell, (b.y - minY) / cell);
+				touched.add(e.from);
+				touched.add(e.to);
 			}
 			rctx.stroke();
 
+			// Member discs for the SEMANTIC era (DECIDED 2026-09-13): clusters
+			// are now topics by meaning, and many members (orphans, isolated
+			// micro-topics) share NO intra-cluster edge with anyone — edge
+			// strokes alone left them completely uncolored. A disc (same
+			// radius as a round cap = dilation) covers exactly the members no
+			// stroke reached: every member is painted by its edges' caps OR
+			// one disc, never both — the no-double-counting rule survives.
+			rctx.fillStyle = `hsla(${hue}, 65%, 60%, ${tuning.edgeAlpha})`;
+			const discR = (tuning.dilation * s) / cell;
+			for (const id of cluster.members) {
+				if (touched.has(id) || !isVisible(id)) continue;
+				const n = this.nodeById.get(id);
+				if (!n) continue;
+				rctx.beginPath();
+				rctx.arc((n.x - minX) / cell, (n.y - minY) / cell, discR, 0, Math.PI * 2);
+				rctx.fill();
+			}
+
 			// Label anchor: member centroid (world coords).
 			for (const id of cluster.members) {
 				if (!isVisible(id)) continue;