feat(mp4): decode QuickTime audio sample entries .mp3, lpcm, twos, and sowt

[nchitkara-xai]

Problem

The QuickTime audio sample entries .mp3, lpcm, twos, and sowt fall
through to UnknownBox, so their sound description fields (channel count,
sample size, sample rate, and the version 1/2 extensions — lpcm entries carry
a version 2 description in practice) are unreachable without hand-parsing.

Fix

Register the four names in both dispatch tables so they decode as
AudioSampleEntryBox, reachable as StsdBox.Mp3 and StsdBox.QtPcm (one
shared pointer for the three PCM names, like AvcX/HvcX — check Type()).
Two safeguards keep hostile or odd files decodable:

• a body that does not parse as a sound sample description becomes an
  UnknownBox with its bytes verbatim, so nothing that decoded before stops
  decoding;
• the StsdBox.AddChild cases for these names use comma-ok assertions to
  tolerate that fallback, so an stsd containing e.g. a truncated .mp3
  entry decodes cleanly instead of panicking.

Note that a .mp3 entry carries no esds; its codec facts live in the sound
description fields themselves.

Tests

All four names decode with correct fixed fields and round-trip; an lpcm
version 2 description populates QuickTimeV2; a truncated body falls back to
UnknownBox byte-verbatim; and an stsd containing that truncated entry
decodes cleanly on both decode paths. go test ./..., go vet, gofmt,
golangci-lint pass.