AudioStreamWAV: Place tag reading code together

[DeeJayLSP]

One thing that bothers me in audio_stream_wav.cpp (which is already messy) is how the code for reading tags is scattered:

1. Fill a hashmap with raw tags found in the file.
2. Near the end of the loading code after all conversions, if tags were found, make another hashmap with "remaps".
3. Use the remaps to fill a dictionary containing the tags.
4. Set the tags.

This changes the code to:

1. Once we know the file has tags (INFO list found), make the remaps.
2. For each tag found, immediately use the remap and put it into the dictionary.
3. Set the tags at the end.

Basically, all tag reading and parsing code is now in one place rather than having parts scattered.

This is one of many PRs I avoided making due to optimism about #96545. I also consider this a requirement for a post-#116184 to add AIFF metadata support, as the current way hardcodes for WAV tags and would need workarounds.