Thai language export to PDF vs Print to PDF, diacritics are garbled

We (the DTP department in a translation agency) also got issues with Thai encoding in PDFs exported from InDesign 2021 (v16).
Just like thread starter Johan we can't use "Print to Adobe PDF" as a workaround because we are working on interactive documents which contain form fields and buttons.
Laubender mentioned that the bug would not exist in INDD2019, but our export to PDF from INDD2019 (v14.0.3) still shows issues, even if not the complete mess we get from the higher versions.

So, here we go, in this post everything refers to INDD2021:

Text encoding is fine in the INDD, compare screenshot 1.
I can properly copy text to my clipboard and place it in a TXT editor. If in the first line one of three Unicode characters gets displayed separately from the rest of the abugida, this seems ok and only got to to with how the editor decides about text placement in this line. In the last line the same three characters ('THAI CHARACTER SO SUA' + vowel mark + tone mark; U+0E2A + U+0E37 + U+0E48) combine well to properly form the Thai abugida in question.
Similar case further to the right, and the rest of the text is completely unconspicuous as far as I can tell.

screenshot 1)

But in the PDF we exported from that very INDD that is no more the case. On the surface it looks fine, but again copying over to a clean text format reveals quite a mess, see screenshot 2.

We now got multiplying tone and vowel marks all over the place (compare the purple boxes), plus empty squares and black diamond question marks which indicate invalid or replacement characters and also some extra empty spaces thrown in between. No good at all.
For a more detailed look, here I chose the first complex (and at once showing a broken encoding) abugida to place on the last line (in the red boxes). First it properly displays the three parts it is supposed to contain: 'THAI CHARACTER RO RUA' + vowel mark + tone mark (U+0E23 + U+0E37 + U+0E48), but then it is followed by unwanted stuff like the characters U+DBC0 and U+DEE0), then adding the same tone mark another time (U+0E48).
Same story with the character we already copied in the first screenshot (orange boxes). After the three correct Unicode characters which already form the abugida (U+0E2A + U+0E37 + U+0E48), we also receive the same unwanted trio (U+DBC0 + U+DEE0 + U+0E48) as above.

screenshot 2)

Our linguists need to be able to properly work with the text in the PDFs, that is to copy the text in its proper encoding over to comment boxes and other tools. We also want to deliver properly working PDFs to our client. Obviously with these PDFs we can't do that.

I will vote in the uservoice post provided by Johan and soon add some more details regarding the issues we see in Thai-PDFs exported from INDD2019 .