Now you have a constant flow. I assume you will pass this flow over/around the cell, and the temperature rise of the exit stream will be the cell output. The ouput signal (hot air temperature) will be pretty warm. If you assume a 100 watt output, then you could estimate the outlet temperature as a guide to choosing construction materials. You can calibrate with internal adjustable IR heaters. This is simiilar to the present air flow device, but I think they do not flow control, but use a dynamic power addition for the calibration. A combination of both strategies would be even better. Use a constant flow, and add a watt or two every once in a while. The outlet temperature units would then be "degrees/watt".
Toasting constantan wire could give many products, depending on the recipe. Do you want to know what your're cooking? There are thousands of possible varaitions to the recipe. Will each batch need to be activity tested until the perfect path is discovered? I think the search surface needs to be broken down into various regions. A wire could be toasted and tested in each of the regions. When some results are optained, intensive search could be more focused. Some possible areas of interest: with bare wire as a starting point
oxidizing environent
low temp
medium temp
high temp
reducing
low
medium
high
cycling reducing and oxidising
low, medium and high temperatures
electroplating extra nickel onto wire for better structual control
(there is list of journal recipes in another post)
Understanding the recipe is important for the next step. We will need to reproduce, tweak the recipe to optimize output (explore the response surface), and scale-up.

Can a rapid test be created? This would allow better experimntal output.

Diodes are great temeprature sensors. Platinum is also okay. Digikey has 1000 ohm (+/- 1%) surface mount platinum temperature sensing resistors for about a dollar .

jdk