makeClientMetadata truncates the appname using Buffer.subarray() and buffer.toString('utf8'):

  Buffer
    .from(appName, 'utf8')
    .subarray(0, 128).toString('utf8')

This breaks when encoding a string that has a multi-byte unicode character that straddles the 128th byte in the array because `toString('utf8')' replaces the invalid bytes with the unicode replacement character:

Buffer.byteLength(
  Buffer
    .from('€'.repeat(300), 'utf8')
    .subarray(0, 128)
    .toString('utf8')) // 129 

Acceptance Criteria

Fix our string length truncation so that we never produce a string that's > 128 bytes.