I will let the video speak for itself and not derail the OPs thread.
Well first, since shading is the precise topic of this thread, I don’t anything about our exchange is the least bit OT.
And second, I went ahead and watched the relevant part of the first video and unless I’m missing something, it agrees 100% with my earlier analysis (for long series strings of conventional panels).
Blindly using analysis and results from ~600V strings to extrapolate what you can expect with shorter string lengths of 2-4S would be foolish.
String inverters fed by long series strings can perform as well as microinverters (and even a bit better since conversion efficiency at high voltage is easily a bit higher than conversion efficiency at lower voltage of Vmp that Microinverters are limited to).
String inverters fed by short (2-4S) strings of half-cut panels will perform worse than Microinverters fed by half-cut panels under almost any shading scenario.
Of course, how much worse is a question of the precise shading situation involved, but I provided a single-leaf shading scenario where a Micro-inverter-based system will deliver 5 to 6% more output.
If you want a more extreme example, consider a string inverter fed by 16 half-cut panels where half the panels are partially blocked by shade across their lower half.
The string inverter can either operate at full Vmp and half Imp or at half Imp and full Vmp, only delivering about 50% of unshaded output in either case.
Hook those same 16 panels half-cut panels up to Microinverters and output will be full Vmp x Imp for the 8 unshaded panels and full Vmp x half Imp for the 8 partially-shaded panels, meaning 75% of unshaded output.
So in that specific shading situation, Microinverters can deliver 50% more power output than string inverters with a long series string.
Half-cut panels have completely changed the math regarding output levels in the presence of shading (at least for certain shading situations which are not uncommon). That was the primary motivation for their introduction.
Referring to old videos and results based on conventional panels which neglect to compare against similar results from half cut panels (often because they did not exist yet) is slightly off point (at least for the OP’s questions).