Some good points here, thank you. Original measurements were done on 30V/3A lab power supply with 3 digit panel meters for current and voltage, but I forgot to consider the leads.
Reference measurement was performed:
CAT6 pure copper wire, exactly 10ft. Untwisting 1m yielded 1.04m, so exact length of twisted pair was 2 x 10ft x 0.305 x 1.04 = 6.34m
The diameter measured 0.52mm, the area is 0.212 mm2
Measured 1.67V / 3.10A = 0.544 ohm. Resistivity = 0.544 ohm * 0.212e-6 m2 / 6.34 m = 1.82e-8
The result is 3.5% higher, which can be explained by increased temperature. Wire was distinctively warm, alpha for copper is 3.9e-3/C, so 20 degrees adjustment brings the resistivity dead on.
The voltage measurements were done with generic XL830L
Checked my power supply panel meters, current was within 0.5% from DMM, the 5.0V was 4.89 on DMM, correction -2.2%
Also during the reference measurement above, power supply leads were determined to be 0.052 Ohm (1.83V - 1.67V) / 3.10A, correction to original measurement 4.76 ohm -1.1%
The original measurement was 74% higher than expected so even with corrections above and accounting about 4-5% for temperature increase, it is still way off.
It is not attracted by the magnet.
Conclusion: mystery alloy
On practical side, I have been continuously dissipating 12W/m entire day today using this mystery wire. Not so great for 1Gbps networking, but works well for de-icing