Generally the point with divided development (note I'm specifically calling this divided development rather than 2-bath development since there is development in both the A and B solutions) is to limit contrast by letting development go more or less to completion in the second bath, which is intended to generate reasonable film speed while reducing contrast. It also typically produces a slightly sharper and grainier negative. D-23 is the basis for most of these developers because of the relatively high concentration of Metol.

The issues I see with what you're suggesting are (1) Highly dilute D-23 is a very slow-working developer which means not much development will take place in bath A unless time is extended considerably and, more importantly (2) not much density will be added in bath B because the developer in the emulsion is so dilute it will exhaust very quickly in bath B. While this should produce a sharp negative, I would expect it to be exceedingly low in contrast with poor speed.

One way to improve things slightly in this scenario would be to use a stronger alkali such as Carbonate in bath B. Since your bath A would be a dilute Metol developer and bath B would be a stronger alkali, notice what you'd essentially have is a type of divided FX-1 developer. But that type of high acutance developer is highly compensating in the first place, so dividing would only result in less speed and even lower contrast. In addition, according to Troop you would have to add a either a significant amount of Sulfite or Sulfite+Sulfate to bath B to control swelling.