Hi, Harry.
divide a flat and apply a bias calibration
Yes. I will. But note: in your case dark=bias.
Also, I think about fast (real time) switch between calibrated / not calibrated view. But it's required twice volume of memory...
Now, if you choose dark1, then dark2,... and again choose dark1, you will every time wait until the script recalculate all Light.
Current method don't requite additional RAM(and I use Redo/Undo to fastest method of reopen original Light), but impossible real-time switching between two dark
So, if we prefer to see real-time results compare of applying different bias+dark+flat+flatdark it's require huge of RAM.
Also, what about the bit depth of files after the calibration? Now, the bit depth before = after calibration. I prefer minimum to increase it twice... requite additional RAM if we like to see "real" result of calibration. I say "real" (not real-time), because I can add to the script one trick: I can decrease bit depth of Light (file in memory) to screen resolution. In that way 1) we must forgot about STF and use only "slow" HistogramTransformation. I mean "slow", because STF need to apply to one frame, HT need to apply to every Light. 2) Statistics require original files (without HT and in full bit depth), so will work from HDD - it's will slow.
So I don't sure about usability of the script...
Yes I know more beer required 
Thanks. No need more, Niall brought enough beer.
