I was originally about to answer the counterevidence... But wow, you are actually correct! I don't have a copy for ISO 8601-1:2019, but 8601-1:2016 [1] has the following note in the section 4.3.2 Complete representations:
> By mutual agreement of the partners in information interchange, the character [T] may be omitted in applications where there is no risk of confusing a date and time of day representation with others defined in this International Standard.
I thought this allows a replacement of [T] with a space, but it doesn't state so! It just means that the time designator can be omitted, e.g. 2023-09-0117:12:30 instead of 2023-09-01T17:12:30. It does look very strange (especially when used with extended formats) and I'm not sure if it was actually intended or not as there is no accompanying example, but nevertheless the 2016 spec doesn't allow a space. Oops.
EDIT: It indeed seems that ISO 8601-1:2019 no longer has this note. I now agree that this is a bad move. https://stackoverflow.com/a/9532375
[1] Specifically the Draft International Standard that was once available from loc.gov.
"8601-1:2016" is not a thing. As you note there was a draft released in 2016 but was changed quite significantly before publication in 2019. That's why the LoC link isn't all that useful for settling disputes.
You are correct that previous editions allowed omission of the 'T' in DateTime expressions. ISO 8601:2004 (the most recent version before ISO 8601-1:2019) states in § 4.3.2:
> NOTE By mutual agreement of the partners in information interchange, the character [T] may be omitted in
applications where there is no risk of confusing a date and time of day representation with others defined in this
International Standard.
This was removed with the 2019 version. However there is another section in the latest version which some people get caught out by. ISO 8601-1:2019 § 5.3.5 states:
> In time-only expressions, UTC of day expressions and time of day with time shift expressions, the
time designator [“T”] may be omitted in the representations defined in 5.3 only when there is no risk
of confusion.
This only refers to Time expressions (not DateTime) and says that both "T16:40" and "16:40" are valid time representations.
> By mutual agreement of the partners in information interchange, the character [T] may be omitted in applications where there is no risk of confusing a date and time of day representation with others defined in this International Standard.
I thought this allows a replacement of [T] with a space, but it doesn't state so! It just means that the time designator can be omitted, e.g. 2023-09-0117:12:30 instead of 2023-09-01T17:12:30. It does look very strange (especially when used with extended formats) and I'm not sure if it was actually intended or not as there is no accompanying example, but nevertheless the 2016 spec doesn't allow a space. Oops.
EDIT: It indeed seems that ISO 8601-1:2019 no longer has this note. I now agree that this is a bad move. https://stackoverflow.com/a/9532375
[1] Specifically the Draft International Standard that was once available from loc.gov.