Incorrect automatic ALSA card ID when unicode is at play

[Barnabás Pőcze <pobrn@xxxxxxxxxxxxxx>]

Hi


When `snd_card::id` is not specified explicitly it is determined automatically
in `sound/core/init.c:snd_card_register()` as follows:

  if (*card->id) {
    // ...
  } else {
    /* create an id from either shortname or longname */
    const char *src;

    src = *card->shortname ? card->shortname : card->longname;
    snd_card_set_id_no_lock(card, src,
                                  retrieve_id_from_card_name(src));
  }

However, `snd_card_set_id_no_lock()`, or more specifically the `copy_valid_id_string()`
function that it calls does not seem to copy very well with utf-8.

For example, based on the report at https://gitlab.freedesktop.org/pipewire/pipewire/-/issues/4135
where `card->shortname="Redmi 电脑音箱"`, `card->id` will be set to \xE7\xE8\xE9\xE7,
which are the first bytes of the symbols in the suffix "电脑音箱" because only those
bytes of the string satisfy the `isalnum()` check in `copy_valid_id_string()`.

I am not sure what kind of 8-bit character set `isalnum()` is basing these results on, but
it certainly does not mix well with utf-8 in this scenario at least.


Regards,
Barnabás Pőcze