If you don't immediately see how the paradoxes Reilly presents in the YouTube videos apply to traditional separate runoff elections, you only need to walk through the exact scenarios presented, but imagine that the IRV rankings are merely preferences inside voters' heads. These voters can vote for their first choice in the initial round of voting. Then in the separate runoff they use the preference order in their heads to decide which of the finalists they will vote for in the second round of voting. Whether the issue is non-monotonicity, winner-turn-loser, participation, etc.... ALL of these paradoxes apply to ALL runoff election systems, not just IRV.
However, IRV does REDUCE the risk of these paradoxes playing any role in an election compared to traditional runoff elections. This is because, in a traditional runoff system, a voter can vote strategically in the first round for a weak opponent, and then switch his or her first choice to the true first choice in the runoff round. Since IRV uses a single ballot and a single round of voting, this trick can't be used. So with IRV attempting to exploit these paradoxes to manipulate the election is far more likely to back-fire than with traditional separate runoffs.