The Heading to Altitude legs are defined in the database as MSL altitudes, therefore the IFD requires baro-corrected altitude to properly sequence.

Reading the available information on the SAE 5-35 Altitude Encoder, I don’t believe it will suffice. Per the Sandia Installation Manual:

The data output of the SAE5-35 is referenced to 29.92 inch HG (1013 Millibars). The SAE5-35 has been designed to provide altitude data to GPS and Terrain Awareness Systems in addition to Mode C Transponders.

It doesn’t appear there is a way to enter or apply a local barometric correction to the output of the SAE 5-35 and therefore it is not sufficient for those leg types.

Another Sandia airdata computer that does support baro-corrected altitude reporting is the SAC 7-35, though it is quite a bit more expensive ($2795 vs $450). Per the 7-35 IM:

If the aircraft has an altimeter with a 5 VDC baro pot, the SAC 7-35 can use this to provide corrected altitude output on the ARINC 429 bus. The Garmin-Z output format does not support Baro Correction information. But, when the baro input is available the data will be included in the King-D and Garmin-G serial RS-232 airdata outputs.

You're right that the IFDs use GPS altitude for the final approach segment of SBAS approaches, but the final approach segment is inherently GPS based.  Think of the final approach segment as you would an ILS beam.  Guidance is provided to the beam, not the legs you see on the flight plan.  Also, The FMS doesn't use DA to determine when to go missed.  It only activates the first leg of the missed approach by crossing the missed approach point - a lateral measurement - just like any other fix terminated leg. Once you're on the missed, you are off of the final approach segment and back on leg-based guidance.  Altitude terminated legs are *always* climb legs, so you'll never see them on the front side of an approach; only the missed. 

Simpson was correct that altitude terminated legs are defined using an MSL reference.  While we can estimate MSL using GPS altitude and a geoid model, we just don't feel comfortable sequencing an altitude terminated leg based on that estimate.  This is especially true of the first leg of a missed approach, which typically takes you to 400 AGL to get you clear of obstructions around the airport.  The last thing we would want to do is sequence early, indicate that it's OK to turn, and then direct someone fly into an obstruction.  I can't say that I know of a regulation saying "thou shalt use only baro altitude" to sequence an altitude terminated leg, but from a safety perspective, we don't feel comfortable using anything else.

I wish we had a more convenient way for you to manually sequence the leg, but there just isn't the real estate available.  On R9 systems, we had a dedicated button for this kind of thing on the PFD.  Of course, in that system, this particular issue wasn't a problem since we have an ADC.  Nevertheless, on the IFDs we had limited choices and the best of those choices was to overload the button on the plan page.