So what do you think the settings should be. 

What I see is that 5.1.3.6.3.6 talks about the designed capacity of the dryers units (regardless of the technology used), it should be capable of delivery air at a dew point below 32 degree F in order to be used in a medical air system. This is different from the dew pint alarm set point, the medical air should be below 35 degree F to comply with the USP requirements, that is  why the alarm need to be set at that level.

It is like the design requirements for the AI compressor, it should be capable to delivery air at a pressure not below 200 psi, but the standard pressure of the AI is between 160 and 185 psi.

This design capacity is determined to this level to allow the refrigerant dryers to be used, in Canada the standard Z7396.1-09 change this design capacity years ago so no refrigerant dryer is allow to be used  

5.6.3.7.1*

Drying and purification units shall dry air to a dew point equivalent to at least –32 °C (–25.6°F) at

350 kPa (50 psi). See Table D.4. Refrigerant dryers shall not be used.

Al, et al ,

You have it right absolutely.  Remember where we start from : Set an objective (be it patient safety (as here) or other worthy objective) and then the standard must comply with NFPA's firm determination that if a method can acheive the desired result, it must be permitted. 

In this case, our desired result is no liquid water.  Can a refrigerant dryer acheive that?  Of course - but only with some caveats.  Therefore, the standard cannot be written to exclude refrigerant dryers.  However, it can be written to deal with the caveats. 

First Caveat: a 'frige dryer can only get to 35 degrees pDp (pressure dew point).  This is because at 32 degrees, they ice up.  The 3 degrees allows for the vagaries of the refrigerant cycle.  So, the published pDp from a 'fridge dryer at compressor pressure is expected to be 35 degrees. Note the standard says design dewpoint of 32 degrees.  That is the Dalton's law effect already accounted for (well, almost - at the 100 psi at which dryers are rated the Dalton's law effect can in theory get you to about 29-30 degrees.  At the more usual pressure that compressors are operated, 80-90 psi, 32 is about as good as you can expect). 

Second Caveat : Experience shows 'fridge dryers have difficulties delivering a consistent dew point across the full flow range, but some manufacturers insist this is not true for their units.  Fine, so the standard says "at any level of demand". 

So the standard also has the alarm set point at 35 degrees (which is as far as the dew point should ever go "off spec" in a properly functioning system using a 'fridge dryer), but a design dew point of "32 degrees", which accounts for the Dalton's Law effect using a 'properly designed system using a 'fridge dryer. 

There is also a small thermodyamic benefit in being below frost point.  Ice requires extra heat to melt to water, and this can help hold water out of the system too.  In our systems, you probably will never see this at work, but it is the reason for the  requirement being expressed as "below frost point" instead of a simple "dew point point of...."