Can you safely review this Digital Isolator?

In my opinion, it can be reviewed safely as long as the scope of the testing is clearly defined.

In a RoadTest environment, I would not attempt to validate the full isolation rating (kV-level testing),
since that requires specialised high-voltage equipment and controlled safety procedures. I would
treat the isolation rating as a specification to trust from the manufacturer rather than something to
validate in a home lab environment. Instead, I would focus on signal integrity, propagation delay,
channel skew, power consumption, noise immunity and overall behaviour under realistic operating
conditions.

That would still provide meaningful engineering insight without introducing unnecessary risk.

btw - I have tested isolator delays using inverter generator and comparing output frequencies, but Gough Lui and his extensive lab would be better prepared (especially when max transfer rate is more manageable 50Mb/s)

Just looking over the data sheet, there are a lot of aspects of this device that would be tricky to test.

Not only the voltage levels, but there are also the timing issues to consider.

This would be a great part for Gough Lui to test.

He would have the knowledge and probably the equipment to do it justice.

Ah, the Megger.  After working in the power industry for almost 40 years, I think the megger gets way more credit than it deserves.  As a DC device, it can create the illusion that the insulation system is better than it really is - for AC applications.  And yet we use it on AC systems all the time.  It essentially measures the very best layer of insulation.  It can confirm if the insulation system is failed but it cannot tell you that the insulation is good for AC systems.
Okay... I'll give you that it's better than nothing AND it's more affordable and less dangerous that an AC Hi-Pot.  Still, know that it's not as cool as it will tell you it is.

there are many parameters that can be tested - when I was roadtesting an isolator, I have tested power consumption vs. signal frequency vs. power supply voltage and delay difference between channels (which is an important parameter because it can generate hazards/races)

I picked up a low cost ($60 USD) megger that had switchable test voltages up to 5000V at a previous job. If a roadtest for the digital isolator included a megger like that for testing, then I could see it being interesting. 

That's something I wondered, although it says 5000v can this be adequately tested under that?

But I suppose you could argue "well that's not testing the full capability of the chip"

Not an expert in this area, but can the isolation specification be tested by a battery powered hi-pot tester.  In spite of the high voltage I think its current is so low that it is relatively safe.

Exactly. If they want independent recognition and validation I'd think any UL recognised test house could do this too.

The intended working voltage is under 1000V