I understand that in a tandem layout, the Cp has to be ahead of the
Cg, and the Np behind the Cg.
For a true tandem (without horizontal tail), it makes little sense to talk about an isolated wing. The combined CP will always coincide with CG (in a static flight). Otherwise, as Jan already noted, the airplane will pitch up.
For NP (neutral point), see below.
Is there a rule of thumb how far forward the Cp has to be in front of Cg?, like 17% to 33% of MAC, etc.
Forget about CP. It's useful in aerodynamics (and somewhat in structural design), but for the dynamics and stability analysis only NP matters. CP is very inconvenient, it moves all the time.
All these 'rules of thumb' apply to NP, not CP. In particular, the distance from CG to NP (CG being ahead) in terms of MAC is your 'static stability margin', and is the main contributor to static stability.
The rule is roughly the same regardless of the aerodynamic configuration. Some minor effects will be different (for example, damping will probably be lower), which may require some minor adjustments.
Should Cm vs alpha, still have a negative slope?
For static stability, yes. This is essentially the definition of static stability; how you achieve this slope is another matter.
Is it possible to have a 50% lift split between rear and front wings, just have a higher incidence in front wing, so it stalls first.
Yes. You'll normally need a somewhat smaller front wing for that. Or you can have a swept front wing and straight rear. Note that if you want a stable aircraft, at a higher angle of attack by definition the lift on the rear wing must rise more than on the front one. If you want to keep this AoA (e.g. at a lower speed), you'll need to dump this extra gain via re-trimming.