I do not understand your questions. Your questions are vague. You must show us the schematics or the article which contains the circuit designs.
|since tubes are notorious for having a short life, running them at a small fraction of their rating is appealing so you won't have to change them all the time.
Tubes have about 5000~10000 hours of life. This is pretty long. This should be adequate for 2~3 years of use. This is not short life at all. If you run tubes much higher than the max rating, then tubes will break so we will assume we are smart enough not to do that. Running tubes at higher or lower rating would not matter as long as the tubes are used CORRECTELY. Biasing is essential.
Another thing that is important is how the amp treat the tubes. The life of a tube will be considerably reduced by applying high voltage before filament supply has a change to adequately heat up the plate. This is why on expensive amps, filament supply goes on before high voltage. In some amps, you will have timed delay circuit so that high voltage supply turns on couple minutes afterwards.
|Low-mu means that you have a low gain and you're probably less likely to add distortion.
Mu is transconductance. It is not exactly mean low gain in all cases, but it is true in this case. However, it is wrong to say that low gain always results in lower distortions. This is not true. Perhaps you have mistaken with noise. Noise and distortion is two different things. Some device can have a bit high noise level but low distortions. Some otherway around. Besides, the gain of tube is so low that if you just use low mu tubes, you will end up with very small voltage swing.
|An external supply is also "good" because you will have less transformer interference with the mu-stage.
This should be true for solid state amps. For tube state amps, power supplies tend to be massive and very very large. So generally, people prefer to have PSU in another chasis. Interference isolation is rather secondary in this case. (Note that mu-stage is ... an amp stage; it is upgraded version of shunt regulated push-pull stage. Sorta unrelated subject. ...)
Personally, I have encountered very little transformer interference problems. If the transformer's core is a loop or a pair of loops, then the magnetic field will stay within the core and leak will be relatively small. Electric field is not our concern since the electric field is only in the wires. So it is safe to assume that transformer interference is insignificant unless of course if you place it close to high gain stages. (which is not our case here.)
Many noise problems in tube amps comes from AC voltage interfering with signal stages. For examples, filament supplies are sometimes AC. This is common for dual indirectly heated tubes like 6922. For these tubes, heater wires are shielded from triode. However, the heater supply wires may come close to signal wires. This can cause hums. Many people use shielded filament supplies now.