IMO US hitting three Gorges (ptentially killing 10s of millions) is basically instantly escalating to proportional countervalue (i.e. targetting civilian, not counterforce, targetting military) nuclear retaliation, regardless of PRC no first use. This isn't perfidy spectrum of warcrime.
I think you're talking about US, willing to escalate to mainland attack, specifically strategic targets that support war economy. Nuclear plants being sensation overblown since it's basically jsut another piece of hard power infra. Which BTW very few US strategic planners have actually indicated willingness to do, but also inevitably must since PRC can prosecute TW (and SKR/JP) war completely from mainland.
To which IMO, most also vastly underestimate the effort needed. Reality right now is, the amount of fire power US can surge in region (naval strikes, aviation regional runway access, CONUS long range bombers), is very limited relative to number of PRC strategic targets, and in contested space theatre. To be blunt, PRC mainland is significantly larger (more targets) and capable (less ability to hit targets) than any previous US adversaries. By 1-2 order of magnitude. Most don't grasp this.
For reference the US+co air campaign in Gulf War, where US+co surged 6 carriers and had extremely geographically favourable regional basing to supplement land aviation, conducted ~100,000 sorties in 40 days, on Iraq, a country 20x smaller (realistically 10x since PRC targets are mostly east half of country), with 80x less people (even less aggregate productive/manufacturing ability). And that campian was essentially UNCONTESTED, since IIRC the french who designed Iraqi anti-air network sold out entire system to west. And it was efficient since regional base (CENTAF Saudi) was close enough that US fighters can sortie with minimal refueling.
None of that is true in PRC campaign, distances involved and limited basing US has access to (at least relative to PRC access to their entire military infra), means US unlikely to forward deploy as much aviation, and sorties need midair tanking (possibly multiple times) to deliver weapons, assuming those fighters aren't shot down/destroyed on the ground in the first place. Same with navy - US can throw in all but the effects won't scale proprotionally since US can't actually sustain/replenish surge for more than a few weeks, assuming support assets don't get destroyed themselves when they restock in port. So to summarize PRC is 10x-20x bigger than Iraq, 80x+ more targets, in contested region where PRC has home team advantage and where US has visiting team disadvantage (with regional partners factored in), in manner that US might not even be able to sustain forward posture for more than a few weeks (vs 5 weeks of initial Gulf War campaign). If you just naively scale Iraq air campaign to PRC, it would take US 5+ years to degrade PRC same way it did Iraq.
That's the scale of problem. Granted it's very hand wavy and napkin mathy but it illustrates how gargantuan PRC actually is and how big the challenges has become relative to US military capability that is calibrated to stomp small/medium sized countries. IMO why planners last 10 years have focused on SLOC/energy blockade, because land war in Asian is stupid. But even blockade talk is going to quiet down (and IMO US supporting TW militarily) in a few years when PRC roles out CONUS conventional strike with ICBMs to mutual conventional homeland vunerability. But that's another matter entirely, the TLDR is US game theory on TW going to be very different when they realize 200-300 oil refineries and lng plants and a few F35 assembly plants can significantly degrade CONUS and NATO. The other part of hitting a 100s of smaller targets vs 1 large target that triggers nuclear retaliation is there's more rungs/opportunity to deescalate, which is probably top priority in actual US/PRC war.