Should an inverter run at 100% or less?

Operating at full capacity will shorten its life span.

A grid tie inverter would be expected to run at capacity
much of its life. A pair of Fronius here have run at clipping
capacity for 8 hours straight every sunny day, for 11 years
and counting.

On the other hand, a battery to local load inverter would not
expect to often be at capacity, and would probably be
engineered accordingly. Bruce Roe

Who/What are those sources ?

For starters, assuming you're talking about string inverters, an inverter in residential PV service will usually not run at its rated output if for no other reason than the input from the array will, for a good portion of the time it's operating, be less than required for full inverter output.

I suppose it's possible to construct some uncommon scenario where a very undersized inverter could operate at its rated capacity for a good part of or even for all daylight hours (and clip a lot), or a larger but still undersized inverter operating with a two axis tracking array or some other configuration and could be at or have input in excess of its full output for longer than when operating with a fixed array, but that's not very likely if for no other reason than such methods waste panel capacity or are uneconomical due to construction and/or maintenance costs.

Even if such things were to be that case, an inverter that's properly sized and located should have no problem operating at its max. output as it will probably not be operated at temps. in excess of its design limits.

The biggest killer of an inverter is equipment temperature. Most well-designed inverters account for heat dissipation in various ways either through active (fan based or mechanical) cooling or natural (natural convection) cooling so that excess temp. of the equipment is normally avoided except maybe under upset conditions.

Still, some users are either ignorant of or disregard mfg. requirements about avoiding situations where heat dissipation and the resulting excess equipment temps that result will be common.

My guess is that most premature failures of quality inverter products result from poor equipment sizing decisions and poorer equipment placement decisions that causes overheating and so early failure.

Some intelligent design considerations and decisions are necessary for satisfactory operation and long equipment life.
Or, maybe morons would be wise to limit array STC sizes to 70% of inverter size.

All that said, if I were to do something stupid like undersizing an inverter, I'd probably put some additional mechanical cooling on it if for no other reason than cheap insurance.

Especially if it's a cheap, poorly designed inverter in a system designed by people who don't know their balloon knot from a hole in the ground.
Quality equipment in well-designed systems will have a high probability of meeting their expectations and design goals including service life.

Many thanks for your reply..
All makes perfect sense.

Many thanks to all for your help

You're most welcome but no sense is perfect, including any I may have.
Get informed and rely on facts you learn through self-education through informed, unbiased sources - not the misleading junk that you get from useless B.S. sources.