Oh no! Not *gasp* crosstalk!
Emissions and crosstalk are a fact of life. It happens. But you know what? Cat5e emissions and crosstalk are decibels away from being problematic at 100bT speeds - at ANY cable length.
The effects of the harmonics you paid so much attention to in your *snicker* physics class represent a microscopic FRACTION of the line loss due to emissions and crosstalk.
As you push up to gigabit speeds, things like Near-End Crosstalk (NEXT) and Equal Level Far End Crosstalk at a frequencies higher than 100MHz become much more of a problem. Which is why CAT6 and CAT7 have much more stringent line loss requirements for the connectors.
That's right. Connectors.
I think you're SEVERELY overestimating the SWR in the mismatched impedances. The changes in resistance due to length are virtually nil in the allowed 100m cable length. Most of the reflections and loss are due to imperfections in the cable, variations in the twist rate, which DO increase with the length. But that's going to be, on average, fairly linear with cable length - it's not going to have an eignvector of stable points, as you suggest.
And to top it all of, you're barking up the wrong tree when it comes to noise. The miseries of the RJ-45 connector, under most circumstances, far outstrip the line loss and reflections in-flight.