Cap local E2E workers so motion assertions stay reliable

Several AGENT=off motion tests poll a layer that exists only for its
180-250ms exit animation. At Playwright's default eight workers on a
developer machine the round trips between assertions outlast that
window, so the exact local suite failed a varying set of these tests
while passing serially. Limit local runs to four workers; CI keeps its
default and retries.

Also stop describing GitHub Copilot as the only model provider in the
README data disclosure and the architecture overview.

Part of #181 (slice 5 of 5).

Assistant-model: Claude Opus 5.5
Assistant-workflow: harness-stack (run 5f1c7488-e4d3-40e8-8307-591dbcff5d63)
Assistant-duration: unmeasured converged, estimated 30m
Assistant-verification: Bun format passed: mise exec -- bun run fix
Assistant-verification: Bun ci passed: mise exec -- bun run ci (existing unrelated oxlint warning)
Assistant-verification: Bun typecheck passed: mise exec -- bun run types (all workspaces and E2E)
Assistant-verification: Bun unit passed: mise exec -- bun test (1591 pass, 2 PostgreSQL skips)
Assistant-verification: Playwright E2E passed: exact flock /tmp/chopin-e2e.lock mise exec -- bun run e2e, three consecutive runs 231/231 exit 0, harness case passed each time
Co-authored-by: Alex Lavaee <lavaman131@github.com>