v0.7.3

v0.7.3 — container self-probe + image HEALTHCHECK

v0.7.3 restores the orchestrator readiness signal for the container image. The
runtime image is distroless — no shell, no wget/curl — so an exec-style
container healthcheck (CMD-SHELL wget …) failed before it could probe anything
and reported a permanent false unhealthy; consumers had to drop their
healthcheck entirely.

• wavefront probe --ready — the binary probes itself. The standard
  distroless pattern: the binary GETs its own ops listener's /ready in-process
  and maps the outcome to the healthcheck exit-code contract — 200 → exit 0,
  anything else (non-200, connection refused, timeout) → exit 1, usage error →
  exit 2. The ops address resolves from the same WAVEFRONT_METRICS_ADDR the
  server binds (only that var — probing never fails because serving-only
  required vars are absent), and a wildcard listen host (0.0.0.0, ::) is
  rewritten to loopback for the dial. No new dependencies (#145 closed).

• Built-in image HEALTHCHECK. The Dockerfile declares the exec-form
  healthcheck — HEALTHCHECK --interval=10s --timeout=3s --start-period=5s
  CMD ["/wavefront", "probe", "--ready"] — so compose consumers inherit a
  working readiness signal with no per-deployment configuration. Kubernetes
  users keep pointing an httpGet probe at the ops port (no exec needed). A new
  image-health CI job builds the image, starts it against a generated bundle,
  and asserts container health flips to healthy.

• feat: add a binary self-probe and image HEALTHCHECK (#146)