@rki looks like you found your answer

can confirm variance in results. had somewhat better results after translating Ehler's code manually
(although my implementation felt way too sluggish)

Yes, the variance in results in huge.
I tried all the methods - Dual, Homodyne, Talib, Autocorrelation, Zero Crossings, Phase Accumulation

The results vary significantly with each other and with Zorro. the DominantPeriod can be very different based on what method is chosen.
I can confirm that PhaseAccumulation is closest to what Zorro shows but only with about a 70% correlation