Triode in both cases, and adjust the anode supply voltage
I don't think that the question in this context makes any sense (maybe I implied that there was a simple answer to your question in my previous reply, sorry about that).
Let's call the "gain figure" mu for now. Assuming identical LC tanks and identical plate voltage in both cases, the mu=10 tube will need tighter feedback coupling than the mu=100 tube to get to oscillation This by itself makes the two receiver setups different so it's not an apple to apple comparison any more. And if you use identical coupling in both cases and degrade the mu of the "better tube" by lowering its plate voltage to match the "older tube", you'll get two identically performing tubes (and consequently identical performance).
I'll try to respond to the questions:
The question is, is the maximum gain figure before oscillation the same for both the x10 tube and the x100 tube?
It should be clear from the above statement that the answer is yes in the mu degrading case, and the overall gain of the system must be Av=1 in any ways (Barkhausen criteria), whether achieved with a higher mu tube, or by using tighter coupling.
And if that is the case, would it be more desirable to use the x10 tube because it would require more regeneration to reach the same maximum gain figure and presumably / consequently have greater selectivity?
Regeneration will always be equal - Av=1.
And is there an optimum gain figure for a tube that is to be used in a regenerative stage?
The optimum mu range is quite wide. You can build a great performing regenerative receiver with a modern high-mu tube as well as with a ridiculously unsuitable, low mu korg nutube (see below). The end result will be the same: LC tank Q will approach infinity (which means that all losses inside it are recovered).