oh, 96*100... I didn't see this relation but I still don't understand the reason behind this.

I also read this in the manual:

"Filter functions are normally cumulative and require a long-enough LookBack period before the filter data is stable. As a rule of thumb, allow about 200 bars lookback period for detecting dominant cycle/phase..."

By setting lookback = 200 I'd expect to have 200 bars in all series. If DominantPeriod needs only 100 bars as you stated, shouldn't it be fine with 200?