The first summer in my house, I turned off the tankless and ran free DHW for about 5 months durring a/c season. As soon as I ran of of hot water once, it flipped it back and never touched it again. In the first 14 months of being in the house I used 334 gallons of LP. We have a gas range used almost daily, the gas grill is hooked to it, used about 3 - 5 times a week minimum, a gas Kingsman vent free fireplace that my wife runs a lot durring the winter. (She loves it!) and the two tankless. 334 gallons in 14 months with all that is pretty good I think.
As far condensing in low delta T uses, I'll have to look up the "why" part. All I can tell you now is it does...and will. Seen it too many times. For instance I went to 5 jobs with one contractor to look at failed HX's on tankless heaters. All 5 were hooked up to First Company Air handler units but without the higher flow pumps. All 5 would only get aobut 1.5GPM flow through the tankless. All 5 set at 140* and getting about 120 -115* back. (That's less than 20,000 btu deliverable heat to the house!) All 5 would never satisfy the t-stat durring the cold days (imagine that) and ran a lot. Long run times at low delta t, with the unit basicly stuck on low fire caused condensate on both sides of the HX which ate them up in aobut 13-15 months. And high gas bills.
The way I see it you still have pretty low temp flue gas because you are not firing very much on low flow conditions for long periods of time and a large section of the HX/burner is unfired for long periods of time.
The key here probably isn't the so much the delta-T as it is the sub-122F return water temps and the low firing rate(=low fire-side turbulence). If you were pumping 10gpm through them instead of 1.5gpm at the same delta-T (for a ~100K burn) it's conceivable that condensation droplets on the fins of the copper HX get scrubbed clear by the sheer draft of the flue. (Is THAT how they manage 50F street water without sustaining damage?) Or is it possible that end-of-burn flue purges are also designed to blow condensate off the fins, and that continuous burns don't see enough purge cycles? I don't know, but I'd sure LIKE to know.
At any firing rate or delta-T, if the return water is above the dew point of the exhaust it simply
can't condense on the HX, (but surely will in the flue.) Even with the same delta-T out of the air handler coil, at same pump flow, if you bumped up the output of the tankless to where the minimum return water temp was 125F they'd probably last for a decade or more. Of course they'd still be freezin' their butts off if the heat delivered didn't exceed the heat loss of the house- sounds under-designed on a couple of fronts.
Lowest modulated fire probably isn't the efficiency sweet-spot either. At lowest fire the turbulence on the fire side of the HX is lower resulting in laminar flow of the exhaust gases over the fins, and the stack temp rises (!). As with water-tube heating boilers, highest combustion-efficiency may in fact hit 86-88% when they're going full-blast, but drop to around 82-83% at quarter-output. (Haven't measured it, but I probably will at some point if I can't find that data from other sources.) Below some firing level the low turbulence on the fire side just kills it on efficieincy though. But for DHW apps it's useful to not have the thing flame-out when the water is at a trickle, so some run down under 10KBTU outputs at lowest fire these days. It could be that at 15-20K the combustion efficiency of those units is only 70%, but if you bumped it to 30K it would be a respectable 80 (?).
These folks measured the raw combustion efficiency of a Rinnai in an air-handler/DHW combi system at between 83-86%, but they don't give details on the modulation levels or water temps at which those numbers were measured. The design heat load was 12kW or ~41K, so presumably it was at least 40-60K when the air handler was running, probably a bit higher during heavy DHW loads:
http://dsp-psd.pwgsc.gc.ca/collection_2007/cmhc-schl/nh18-22/NH18-22-106-108E.pdf
Their cousins, the copper-tube hydronic boilers (eg RayPak) seem to tolerate ~110F return water on long burns without damage, (that's with ~10% excess combustion air and low-80s combustion efficiencies) but even the multi-stage versions don't drop below ~25% of full-fire. There must be something to fin-tube HX technology that's at least somewhat tolerant of low temp as long as the fire-side flow/turbulence is high enough.
Tankless units serve radiant slabs at even lower return temps all the time, seemingly without rapid deterioration of the heat exchangers. But I'm wondering if there's some minimum modulation setting at which they need to operate to avoid self-destruction? It would be nice to get to the bottom of this, since tankless heaters are being designed into combi heating systems at an increasing rate. Takagi seems to be marketing their stuff as generic wall-hung boilers (without any low temp or low fire caveats in their literature) but Rinnai has its air-handler combi, and Navien has it's heating-box combi heat exchanger setup too.
From your experience it sounds like the mixer-valve solution for heat pump + tankless in series has very reasonable efficiency, and it's clearly a low-cost low-maintenance no-headache solution. If the cold-water sandwich issue isn't annoying (which it won't be in warm-water areas the way it is with sub-40F mid-winter water in other areas) there's no reason to do anything different.