Why is a single supply voltage is seldom used for audio mixers”
Because in audio circuits the circuits become much simpler when used with a symmetric supply voltage (like +/- 15V). If one were to use a single 30 V supply (for example) then you have to take measures to DC bias the inputs of the opamps to a voltage somewhere "in the middle", with a 30 V supply that would be +15 V.
You could also create a +15 V "AC ground" rail to DC bias all the inputs from but then it is often just a small step to make the power supply deliver +/- 15 V. Then all the inputs can be DC biased to 0 V which is "neat" and "easy". Then the signals will also be centered around 0 V (ground) instead of having +15 V DC added to them.
Designers will often push an op amp up to, and sometimes beyond its recommended voltage rails to increase dynamic range
Does this still apply ?
Yes, in principle it does. Using the maximum supply voltage does indeed give the maximum output voltage swing which is what is needed to reach the largest dynamic range. Feed an opamp with +/- 5 V and even the best opamp is limited to 10 Vpp output voltage swing. Use +/-15 V and you get 30 Vpp output voltage swing.
However, not using the recommended supply voltage (but instead, using the maximum) might result in issues which might not be listed in the datasheet. But usually there should be no issue unless you'd exceed the maximum.
All circuits have a limited dynamic range, depending on the opamp you get more range or you get less.
You cannot say that the number of channels is limited by the dynamic range. As long as you take care that the summed signal (current) is below what the circuit can handle (the output of opamp will not clip) then there is basically no limit on the number of channels.