If instead of g/mol, we referred to molecules/g -- we would end up populating tables and charts with really big numbers. This would make lookup tables hard to read, difficult to publish, and hard to work with. Imagine if you had to do math with a bunch of 10^23 exponents all of the time.
Instead, it was agreed to effectively pull out a constant value from each of those to make the math significantly easier. Now, instead of dealing with a lot of big numbers, all of the lookup tables could now list smaller g/mol values. And we would be left with just the one single large (Avogadro's) number in the equations.
Honestly, we don't need a set mole constant, but it makes chemistry significantly easier to do so. Unlike the other constants mentioned in the OP, Avogadro's number is completely arbitrary. It could be '1' as the parent suggested, except then it makes the rest of the math more difficult.
Even for this SI overhaul, we didn't really even need to redefine the mole, except for the fact that it was previously defined in terms of the old kg. This was just "fixing a glitch".