do you label transformers? How about transformers less than 125kva with 208/120Y secondary. Are they excluded along with the panelboard they feed?
Can you set an incident energy less than 1.2 to create a label or use the calculated value?

The upstream side can still be a hazard. If both primary and secondary are in the same compartment, that's an issue. If they are separate then maybe only one side needs to be labelled. But if you don't intend on accessing it live then a label may not be required at all as per NEC. On oil filled transformers particularly padmounts where the oil sample and instrument gauges are inside the door, a label is needed. In a lot of dry transformers a label may not be required at all.

^^^^ This ^^^^^

Dry types - no.

Oil filled pad mounts - yes.

Incidentally, I recently went to a sales pitch disguised as a learning session about at least one of the pad mount guys (Eaton/Cooper) offering VFI switches on the pad mount primary with CT's on the secondary so the primary side could be tripped from the CT's thereby reducing incident energy on the pad mount secondary enough to allow opening the secondary door while energized with appropriate PPE. This will also reduce the energy at the line side of the MDP main. I think I am going to specify that feature for the next pad mount I buy. I usually buy 1 or 2 pad mounts or so per year for the university I work at. We already went to exterior lockable oil drain valves and sampling ports a couple years ago so we could take oil samples for DGA while energized and not have to open the secondary door.

Good plan. The VFI is a bit pricey and there are other switches/breakers on the market such as the ABB magnetic breaker, and PDP has repackaged it with an integral disconnect switch for a really nice compact setup. I've used Tavrida's mounted on stands and you can really use anyone's recloser if you are already distributing over a pole. And there are a couple really well made underground breakers that work excellent, too. I've driven them with an SEL-651. This is really intended as a recloser type of protective relay but the big advantage is that this single relay has 6 CT inputs on it and comes with the battery and other hardware needed for essentially a self-contained unit that is already done for you if you buy it that way. Just have to turn all the reclosing stuff off and change the labels to be a little more human readable.

But here's another interesting version. I had a semi-remote pumping station with two transformers feeding two medium voltage MCC lineups from a single feeder. Either transformer was capable of feeding both MCC's. So when I went in to rebuild/upgrade the delapidated feeder breakers, I realized I could get primary breakers cheaper than secondaries. So I bought two primary side breakers to feed each of the transformers with primary and secondary CT's in the so-called "virtual breaker" arrangement. Then on the secondary side I bought 3 switches with trapped key interlocks with two keys so that only two of the 3 switches could be closed at a time. The manual switches were arranged in a main-tie-main arrangement. So under normal operation the two transformers fed two switches which fed the two MCC's. The switches would be closed and the third switch (tie) was open. During transformer maintenance, one breaker and one switch could be opened while the tie would be closed, isolating the transformer. The switches were vastly cheaper than the same arrangement would have been with 3 breakers and say two primary side fused disconnects and in addition the arc flash rating was obviously very low for the entire arrangement, particularly since the tie could not be closed with both transformers energized.

Using primary devices for secondary protection may cause a challenge with limiting arc flash exposure, as the primary device needs to support magnetizing current. In one location we were able to achieve desired results by implementing a 'time delay instantaneous' function which seemed to cover this gap. This scheme was successful without secondary CT's connected to the primary device.

Ordinarily I'd agree. Even then it's hard to get enough selectivity from the primary side to do much in the way of secondary side protection in many cases. But there's another option. Let's say for a moment we're doing breaker-based protection and we do "full" protection so we have breaker-transformer-breaker. With this configuration we have arc flash protection on the primary side of the transformer and we have a zone that is relatively unprotected from the transformer until it reaches the secondary breaker. So let's fix that. We'll still use a protective relay but we'll mount bushing CT's directly on the transformer secondary terminals and we'll trip the primary side breaker, too. With this configuration, the secondary side breaker becomes redundant and can be eliminated. This leaves us with two different protection relays tripping the primary side breaker. As a bonus the unprotected zone almost disappears. Now let's develop this one step further. If we take into account the transformer ratio, add restraint to account for accuracy limitations, and use the 3 CT's on the primary side and the 3 CT's on the secondary side in a differential relaying (87 relaying) scheme, now we have full 100% protection for the transformer. We can still use the primary and secondary side CT's for overcurent protection. In a digital multifunction relay we can do 50/51/87, and 50/51N relaying so that we have full protection of both the transformer and downstream devices.

This is vastly different from trying to do primary-side only protection which if as you suggest we use an auxiliary contact on the breaker/switch or run the trip/close functions through the relay itself then we can use this to restrain the overcurrent protection to avoid the inrush. This reduces the coordination needed so that the overcurrent protection can be set lower but it is nowhere near the selectivity of monitoring on both sides of the transformer.