Suitability of Two Power Supply

I'd advise you to keep the separate supplies - you can get problems with noise from switching on the outputs affecting your input side.

10A sounds an awful lot of current for the inputs - maybe you could make a saving by checking how much capacity you really need for inputs, and fitting a smaller supply.

Another thought is to go over to 24Vac on the output side (or, if changing over to ac is too disruptive, use a simple unregulated supply - transformer/rectifier/smoothing cap), either of which would be cheaper than a regulated supply.

Yes you can use one 20 amp supply then split it in to the capacities required. I would suggest use of a rectifier bridge or UPS to supply (contactors,solenoids and relays) to eliminate noise and for protection. Use of DC is most efficient.

You have to consider two things: what happens, when a power supply fails, and what happens when (not if) a noise from the power section causes a glitch in the logic section. If the answer is - as it should be - messed up production and broken CNC machine, the answer is clear. Two separate power supplies are cheap insurance in this context. Additionally, a shop using a number of your machines have to stock only one type power supply.

I disagree that two are better than one. With as single power supply powering the entire system, if it fails, everything stops. With two power supplies, a careful analysis is required to determine what continues to operate and what the consequences would be. Suppress the inductive kick by placing a movistor or diode across each coil and you should be ok. Another thing that you should consider is distributing power and common to each component using a star topology and twisted pair wiring to improve immunity to EMI.

A relay can be added across the output of the power supply for the logic circuits with normally open contacts in the output of the supply to the sloenoids and relays. Therefore, it the logics supply fails, it cuts power to the solenoids/relays thus preventing uncontrolled actions.

Joe is correct on the interlock for a proper shutdown and startup. Snave is correct as far as he went with suppression and some such. But, what I said and find necessary to emphasize is, that switching power - 24VDC relays, or 220 or maybe 480VAC - physically near to the logic causes unavoidable bounces on power, logic and ground lines plenty to upset the logic. Then whatever happens, happens. I have not met anybody, including myself (pun intended), who could overcome these problems. The proper way is isolating logic from the power section, with proper sequencing and interlocks. The big name chip makers, like National are full of Notes to this effect.