My bad, I was served this screenshot as the default. From the URL, I see it was wrong.

Does macOS offer a default sound conversion application?

What app is that?

What will the Apple Music app do, when you request conversion to AIFF like this? https://support.apple.com/en-us/108961

Are you referring to Apple Music not supporting CAF

• The ones with "sowt" and NONE in the filenames, are generated by flac. I don't know what applications do read "sowt" that aren't 16 bits. Cf. https://hydrogenaud.io/index.php/topic,127175.msg1059384.html#msg1059384 .
• The ffmpegs are generated with -acodec copy; for unsigned 8-bits that results in type "raw" which I have not found generated elsewhere, but which is a type described for QuickTime, see previous link. WavPack usually does things right, but does not know this one. I suspect there is nothing such in AIFF really, and that ffmpeg devs have just taken the QuickTime types because, if user forces -codec copy then do codec copy and get as close as you can.
• And a couple of ALACs for good measure.

• fb2k reads the 8-bit "NONE" and "sowt". IIRC 8-bit "sowt" support was introduced quite recently - and I don't even know if it is supposed to be a thing. And ffmpeg evaluates them to identical MD5 (which do agree with FLAC and WavPack, all OK here.)
• fb2k also reads the 8-bit "raw" that ffmpeg creates, and bitcompares it to identical to the "NONE" and "sowt". Edit: The reason why ffmpeg evaluates that "raw" to a different MD5, is because it is unsigned, like .wav - it returns same MD5 as for the corresponding .wav. This is because ffmpeg - like WavPack - evaluates MD5 in source's endianness and signedness. FLAC on the other hand, converts everything to signed little-endian.
  
• 16-bit files are all read and all OK. Two MD5s are produced because FLAC evaluates MD5 as if little-endian, while ffmpeg and WavPack evaluate in the source file's byte order.
• 24-bit: 

... I should likely have included both AIFC sources then, to see if those are read in all thinkable flavours. Attached. Which of these work?

]$ sndfile-info atest1ch-24_0.ffmpeg.aifc
Error : Not able to open input file atest1ch-24_0.ffmpeg.aifc.
File : atest1ch-24_0.ffmpeg.aifc
Length : 31824
FORM : 31816
 AIFC
 FVER : 4
 COMM : 24
  Sample Rate : 44100
  Frames      : 10584
  Channels    : 1
  Sample Size : 24
AIFC : Unimplemented format :

File contains data in an unimplemented format.

• either FLAC writes wrong 8-bit sowt
• or Apple Music cannot read 8-bit sowt properly
• or possibly: 8-bit sowt should not exist and therefore there is no "right" way to read nor write them

As for ffmpeg, it is easier to conclude that ffmpeg does wrong things, when it writes files it cannot read.

     { AV_CODEC_ID_PCM_S16BE,    MKTAG('t','w','o','s') },
     { AV_CODEC_ID_PCM_S16LE,    MKTAG('s','o','w','t') },
+    { AV_CODEC_ID_PCM_S24LE,    MKTAG('s','o','w','t') },
     { AV_CODEC_ID_ADPCM_IMA_QT, MKTAG('i','m','a','4') },
     { AV_CODEC_ID_QDMC,         MKTAG('Q','D','M','C') },

const AVCodecTag ff_codec_aiff_tags[] = {
    { AV_CODEC_ID_PCM_S16BE,    MKTAG('N','O','N','E') },
    { AV_CODEC_ID_PCM_S8,       MKTAG('N','O','N','E') },
    { AV_CODEC_ID_PCM_U8,       MKTAG('r','a','w',' ') },
    { AV_CODEC_ID_PCM_S24BE,    MKTAG('N','O','N','E') },
    { AV_CODEC_ID_PCM_S32BE,    MKTAG('N','O','N','E') },
...
    { AV_CODEC_ID_PCM_S16LE,    MKTAG('s','o','w','t') },
    { AV_CODEC_ID_PCM_S24LE,    MKTAG('s','o','w','t') },    // <--- added by me

        par->codec_id  = ff_codec_get_id(ff_codec_aiff_tags, par->codec_tag);    // <--- A
...
    if (version != AIFF_C_VERSION1 || par->codec_id == AV_CODEC_ID_PCM_S16BE) {
        par->codec_id = aiff_codec_get_id(par->bits_per_coded_sample);           // <--- B
        par->bits_per_coded_sample = av_get_bits_per_sample(par->codec_id);

Apple Music then reads such file but treats it as 16-bit.