Dan has just about covered this perfectly.
Tsr has to be the design value and it will be the peak of a series of power curves, it is the optimum value. You will have curves very similar on either side of the design figure. If you go far too low you hit stall and power( Lift) drops like a stone ( much like it does when a plane wing stalls).
When unloaded there is only drag to limit speed, some profiles will reach nearly 3 times the loaded value and some will become too lossy to go up to double loaded speed.
The calculators only give you results that work at constant tsr. If you load in such a way that you don't track the peak of the power curve you may find that the large chord towards the centre and variable pitch are actually worse. For conventional loading I haven't found the calculator predictions helpful and I tend to use the 70% radius figure outwards and extrapolate this linearly to the centre ( very similar to Dan's blade designs) and it works better and uses less timber.
I don't have enough experimental data with mppt to know if following the calculator predictions works better. The machine I have it running on has a prop with twist that reasonably follows the calculator figures to about 1/3 radius, I seriously doubt that going to the extremes near the centre would be beneficial.
One thing is certain you will gain more from mppt than any amount of clever aerofoils but you may possibly be able to get something a bit better with the best available profiles once you have got the tracking to the power curve right. Without mppt don't waste time on even trying to copy accurate profiles and I also suspect that with mppt you will have to copy very accurately with accurate templates of profile milling to get any measurable difference over a sensible hand carved respectable blade.
My own experience is that the optimum tsr falls with wind speed and I get best results by setting my mppt to increase load beyond the theoretical for the higher wind speeds.
This may be something where you could get an improvement with better profiles but it may also be a fact of life and it is certainly fortunate for those who run conventional loading as it at least gives them a head start against the stall problem and may be one reason why simple loading sometimes does better than you would expect.
I seem to be able to get an overall Cp of 0.3 over a fair range of wind speed and with the converter and alternator efficiency about 70% this suggests that the prop is doing better than 0.4. I doubt that you will get above 0,45 if you spend the rest of your life building the perfect blade and even then it will loose shape and surface finish from erosion and the clever profiles will suffer very badly compared with the simple hand carved things.
The big problem with all this is that it is virtually impossible to measure these differences without facilities beyond most of us. Just comparing output against an anemometer on a weather station is little more than guessing. Instantaneous reading anemometers mounted just upwind of the prop at centre height and elaborate data logging into bins of wind speed is the very minimum needed to see these differneces.
Perhaps comparing mills at the same height on towers not far apart in clean air may be the best comparison but few of us can hope to do it.
Flux