Align the cap emulation on NextGraph's model, and confine it to a virtual user

Two batches, verified against nextgraph-rs throughout.

P1a — the capability surface. Reading was an ACL (Map<doc, Set<principal>>),
the exact inversion of key possession. It is now possession: `capFor(nuri)` is
the only question, there is no principal parameter anywhere, and nothing turns
a bare reference into a cap. Sharing is `shareCap(cap, toInbox)`, a Link
deposit; receiving needs no operation. `Nuri` and `ReadCap` are template
literal types, so passing a bare reference where a cap belongs is a compile
error, with runtime guards behind it for JavaScript callers.

The virtual user boundary. Every access function is now confined to the
connected user, through two rules on one criterion (possession), implemented in
two places so a lapse in either is caught by the other: authorization at the
passage points, and "do not even attempt" at the callers. The polyfill's own
machinery moved to physical.ts — unguarded, never exported — which replaced an
exemption list: the machinery no longer gets waved through the guard, it calls
something the guard never saw.

Removed, as emulating capabilities the target does not have:
- discovery.ts and its global index. There is no discovery in NextGraph; you
  follow links. It also pooled user data across wallets.
- the cross-account fan-out (listEntityDocs, resolveReadGraphs, allAccounts,
  loadShim), which was cross-user enumeration by construction.
- resolveInboxAnchor, a single inbox common to every user.

Caps are now stored where NextGraph stores them, and read back rather than
recomputed: AddRepo on the store's Store branch for documents a user creates,
AddLink on its User branch for caps received. Inboxes belong to someone — the
user's own, plus one per document — and connecting a user drains them all;
that is the library's job, not the app's.

Corrections worth recording: a ReadCap is `r:`, not `:k:` (reported by
NextGraph's developer, verified in BlockRef::readcap_nuri); received caps DO
have a register (AddLink), contrary to what this repo's notes claimed; and
"wallet" upstream means keyring — what owns three stores is a user, so the
vocabulary follows.

The cap value is the constant OK: the only question the emulation answers is
whether a cap is held. P1b replaces that one constant with a real key.

After this the shape is right and the isolation is still fake. Nothing here may
be described as anonymous or private.

packages/client/test/read-model.test.ts

@@ -1,6 +1,21 @@
-import { test, expect, mock } from "bun:test";
+import { test, expect, mock, afterAll } from "bun:test";
 import { readUnion } from "../src/read-model";
-import { configure, configureStoreRegistry } from "../src/polyfill";
+import type { Nuri } from "../src/types";
+import {
+  configure,
+  configureStoreRegistry,
+  getCaps,
+  resetCaps,
+  setCurrentUser,
+} from "../src/polyfill";
+
+// The cap registry is process-wide, so each inject() starts from an empty one:
+// once ANY cap exists the possession gate is in force for every reader, and a
+// suite that never declares caps must not inherit another suite's.
+afterAll(() => {
+  resetCaps();
+  setCurrentUser(null);
+});
 
 // A fake `ng` whose sparql_query answers the ANCHORED per-doc query (SELECT ?s ?p ?o
 // WHERE { ?s ?p ?o }, anchor = the doc NURI) with ONLY that doc's triples. There is
@@ -31,6 +46,8 @@ function fakeNgWith(triplesByDoc: Record<string, Array<[string, string]>>) {
 
 function inject(triplesByDoc: Record<string, Array<[string, string]>>) {
   const ng = fakeNgWith(triplesByDoc);
+  resetCaps();
+  setCurrentUser(null);
   configure({ ng: ng as any, useShape: (() => {}) as any });
   configureStoreRegistry({
     getSession: async () => ({ sessionId: "sid-rm", privateStoreId: "priv" }),
@@ -103,3 +120,26 @@ test("a doc that fails to read is skipped, not aborting the batch", async () =>
   // The bad doc failed its read but the good one still lists.
   expect(subjects.map((s) => s.subject)).toEqual(["did:ng:o:ok"]);
 });
+
+// The possession gate, at the read-model's own level: once ANY cap exists, a doc
+// whose cap is not in what the current holder holds is dropped — however well its
+// NURI resolves. Before the first cap the gate is inert (no regression).
+test("readUnion drops a doc whose cap the holder does not hold", async () => {
+  inject({
+    "did:ng:o:mine": [[TYPE, `${FP}Event`], [`${FP}title`, "mine"]],
+    "did:ng:o:theirs": [[TYPE, `${FP}Event`], [`${FP}title`, "theirs"]],
+  });
+  const both: Nuri[] = ["did:ng:o:mine", "did:ng:o:theirs"];
+
+  // Inert: no cap issued yet → everything flows through.
+  expect((await readUnion(both)).map((s) => s.subject).sort()).toEqual(both);
+
+  // One cap issued → possession is now the rule for every document.
+  setCurrentUser("alice");
+  getCaps().mint("did:ng:o:mine");
+  expect((await readUnion(both)).map((s) => s.subject)).toEqual(["did:ng:o:mine"]);
+
+  // …and for every holder: bob holds nothing, so bob reads nothing.
+  setCurrentUser("bob");
+  expect(await readUnion(both)).toEqual([]);
+});