0.8.2 — vocabulary 16: reporting fixes for dropped layers, asset ids and signal tracks (#154, #155, #156, #157)

Reporting fixes for dropped layers, asset ids and signal tracks (#154, #155, #156, #157). The
vocabulary and every reference frame are unchanged.

• validateAnimSpec reports layers past the 12-layer cap, at the top level and in each clip
  (layers: only the first 12 layers are used (14 given)), and lists their types in
  dropped (#155).
• animspec render prints the validator's errors as a JSON line on stderr, and a new
  --strict flag exits with code 2 instead of rendering when anything was dropped or
  reported (#154).
• An invalid asset or plugin id now states the rule: lowercase letters, digits and -,
  starting with a letter, at most 16 characters (#154).
• type-reveal: the x and align descriptions state that rows align inside a maxWidth
  box centred on x. The validator reports a left- or right-aligned box that leaves the
  frame (#156). Rendering is unchanged.
• decodeSignalTrack accepts raw file bytes (any byte view) as a track with no sidecar, and
  a wrong argument throws an error that names the expected shape (#157).