ISO core standard says atom directly followed by lpar is always compound.

[ghost]

In the ISO core standard, when one studies the syntax of a compound, one
finds that a special lpar is used. Basically an atom directly followed by an lpar,
should always lead to compound, even if the atom is an operator. Otherwise

write_canonical wouldn't work, you could not read the result back anymore.
Here you see the expected result in GNU Prolog:

[ghost]

Somehow this doesn't work anymore in Tau Prolog. Maybe a result
of the introduction of the (+)/1 operator. Here you see a screenshot,
which if I read it correctly does not give two different results

as GNU Prolog does:

[ghost]

The relevant ISO core standard syntax clause is the following.
This assures that write_canonical can always be read back:

6.3.3 Compound terms - functional notation

term = atom, open ct, arg list, close

It uses the "open ct" non-terminal and not the "open" non-terminal. The
"open ct" non-terminal is without layout text sequence in front. So it says
basically atom directly followed by left parenthesis should give compound.