Good questions Bobrex. I will try to cover some basic info for newbies who may not be electronically inclined.
I think the first question that needs answering is why low output impedance of a preamplifier. (Amps require different treatment.) There are two main reasons, not necessarily in order of importance.
The first is because the interconnect capacitance (IC) and input capacitance of the following device (say amp) tend to reduce the high frequency response. (IC capacitance tends to dominate among the two.) But is a low output Z necessary? Actually a resounding no. One can use low plate resistance tubes, thus no need for a low impedance buffer stage, unless one is using long interconnects.
Secondly, to reduce hum pickup. But again, that is mainly for long IC runs, or for some reason a problem with hum pickup.
Next, we need to discuss if low impedances change the sonics, adding more "slam, impact". Actually no in one sense and yes in another. I have tested alot of capacitors, so if one uses an inferior cap, yes. Since caps sound different, only one or two caps can be the most accurate vs a straight wire, the best will not limit in dynamics vs dc coupling. (Less one is tempted to ask me what caps I use, some are special designed/ordered. I also have not checked every cap in the world.) But many designs use small coupling capacitors, thus the deep bass is limited.
"This also results in deeper lows" (DC coupling)
That is my point. Yes, deeper lows, but not accurate lows. The deeper lows are the result of the lower bass being overly excentuated (more so than the harmonics) because of the inherent power supply design. The result is that the bass, although sounding deeper, is not natural and accurate to the incoming signal.
(DC coupled) "But such circuits, at least to me, also sound artificial and overly lean - they typically make string basses sound - ummm, wrong."
Yep you are probably right. Excentuating the bass usually affects the midrange, or even higher as well. This is because DC couple circuits almost always utilize more power supplies to muck things up. That is one reason I don't design balanced designs, more power supplies
(except in multiple stage components such as amplifiers to prevent feedback).
The question then becomes if an extra power supply or two is better than one coupling cap? (If one designs so as not to need another power supply, some other complicated setup is needed to couple the stages together.) The answer depends on the quality of the coupling capacitor. Most available today are poor, but I would usually say no.
"is there a relationship between the DC coupling and ultra low output Zs?"
Can be. Depends on the output capacitor size. PBN preamps and amps are designed so one can connect a 75 ohm output to 75 ohm input. There, a coupling cap would not be good.
The main reason cited by many manufacturers is that DC coupling will drive the Input of low input Z amps. But unless the input Z of the amp is below 20k ohms (I may be conservative), there is no need for either DC coupling or low output Z from the preamp.
That last statement needs to be qualified. The RCA Radiotron Designers Handbook, 26 engineers and the 'bible' so to speak, recommends the output Z of the preamp be no more than approximately 1/5th that of the input Z of the following stage. So with 20k input Z, the preamps output could be, say 4k output. I recommend less than that, say 2k. But this is only to make sure the HD distortion doesn't rise appreciably. Also, with a 4k output Z, even low capacitance cables can affect the highs. This based on listening tests.
As you mentioned, why add a buffer stage (because of high output Z tubes); another stage that will only deteriorate the sonics further? And why use high capacitance cables that will also deteriorate the sonics? Eliminate the buffer stage and lower unwanted capacitance, thus killing two bad birds.
"Ever notice how some old designs sound so good? One reason is that they may not use a buffer stage, an extra stage that doesn't do much, except lower the fidelity."
Yes. And even buffer stages have harmonic distortion (HD). This means that when the HD of the gainstage passes throught the buffer stage, higher orders of HD are present. It is important to keep the HD in all stages as low as possible.
Hope this helps.