ESD diodes absorb very small amounts of energy, but at high short term energies. A repetitive signal would likely put those short spike at a rate that exceeds there thermal capabilities of the package.
Also your working voltage of an 0805 is commonly about 150V. Which depends allot on which creepage spec you use (pollution degree, etc), the specs vary quite allot. I use the 150V as it's a middle of the road low humidity spec. That's why I use 2X of 0805 package, to get a combined 300V. Technically this could allow for a 600Vp-p signal with 300V isolation to your GND or 5V rail. The concern is that creepage currents can ionize and dump excessive current to your rails. See below PDF for a general working voltage spec.
https://www.bourns.com/pdfs/chpreztr.pdf
Also when thermally cycling a PCB, in an industrial range, you commonly have about a 50% failure rate with 1206 components, and higher failure rates with larger components. The differences in thermal expansion from the PCB to component, commonly causes the copper adhesive to shear off PCB, fatigue cracks, etc. I suggest avoiding 1206 and larger packages because of thermal cycling issues.
Rail clamping diodes don't necessarily start to conduct at .7V from the rail. Those diodes could start clamping at like 15V or more if MAX so choose. I'm not sure what MAX has done. I see this note in the datasheet, which seems to indicate you should run a 300V signal, then measure if your current is above or below 40mA.
Current into IN+, IN-, IN_+, IN_-.......................................±40mA
I'm not sure how this works when it's powered with a 5V rail. Even if that's +/- 2V for 4Vp-p, you still aren't at 5V. I must be interpreting this datasheet incorrectly. Or perhaps they have some magic inside.
Overdrive = 2V to 3V, zero-crossing
Keep in mind you want your 10k resistors to be large enough that you get enough energy at low RPM to accurately detect the signal, but don't have to much energy at high high temperature and high RPM's. If you limit your detectable voltages with external clamping diodes you'll limit the range you can detect, such that it will either be harder at low RPM, or harder at high RPM.
I think the MAX chip is clamping to 5V, perhaps as high as 5.1V. I would use the internal clamping diodes, instead of adding external ones. I could see how an external diode could have a slow recover time which causes it to skew the signal. I could also see how a fast switching diode might create high frequencies that can propagate past the other circuitry on the chip. Fast transients can cause latch up issues, etc. I would suggest that people use the Honeywell datasheet, and Honeywell product line to figure out what kinds of voltages they might expect.