While I agree that the description of operator overloading should be better integrated into [expr], I disagree that the change "precedence" -> "syntax" is helpful here. The sentence you quote is essentially a note (nothing normative would change if it were omitted), and "precedence" is a lot more specific (and thus easier to understand) than "syntax".

While I agree that the description of operator overloading should be better integrated into [expr]

Yes, this is the subject of #3957 and #5315. The relevant issues that can be fixed by the integration might be #5172, #5401, #5208. Integrate [over.oper] into [expr] would make [expr] self-consistent; any operation that forms an expression should be covered by [expr] regardless of what type of the operand is. [over] should only concern the resolution/selection between two functions. If we integrated [over.oper] into [expr.compound], the quoted rule could be changed as a note.