Clear bus_hvdc_net_power in clear_injection_data!

[luke-kiernan]

Zero bus_hvdc_net_power each solve, alongside the other injection fields, so construction-time native HVDC setpoints don't double-count and per-solve writes don't accumulate.

🤖 Generated with Claude Code

[github-actions]

Performance Results

Precompile Time

Main	This Branch	Delta
2.456 s	2.389 s	-2.7%

Solve Time

Polar AC

Test	Main	This Branch	Delta
NewtonRaphsonACPowerFlow First Solve	15.178 s	15.819 s	+4.2%
NewtonRaphsonACPowerFlow Second Solve	78.1 ms	78.7 ms	+0.8%
RobustHomotopyPowerFlow First Solve	15.948 s	16.386 s	+2.7%
RobustHomotopyPowerFlow Second Solve	9.115 s	8.7 s	-4.5%
NewtonRaphsonACPowerFlow(iwamoto) First Solve	194.9 ms	199.3 ms	+2.3%
NewtonRaphsonACPowerFlow(iwamoto) Second Solve	78.6 ms	499.8 ms	+535.9%
TrustRegionACPowerFlow(iwamoto) First Solve	2.227 s	1.69 s	-24.1%
TrustRegionACPowerFlow(iwamoto) Second Solve	78.2 ms	77.8 ms	-0.5%
NewtonRaphsonACPowerFlow First Solve	1.005 s	1.097 s	+9.2%
NewtonRaphsonACPowerFlow Second Solve	1.002 s	1.017 s	+1.4%
TrustRegionACPowerFlow First Solve	1.106 s	1.471 s	+33.0%
TrustRegionACPowerFlow Second Solve	952.9 ms	998.2 ms	+4.8%

Rectangular CI

Test	Main	This Branch	Delta
ACRectangularPowerFlow{NR} First Solve	5.46 s	5.12 s	-6.2%
ACRectangularPowerFlow{NR} Second Solve	40.2 ms	38.3 ms	-4.7%
ACRectangularPowerFlow{NR}(iwamoto) First Solve	160.8 ms	159.4 ms	-0.9%
ACRectangularPowerFlow{NR}(iwamoto) Second Solve	38.2 ms	38.8 ms	+1.5%
ACRectangularPowerFlow{TR} First Solve	5.782 s	5.626 s	-2.7%
ACRectangularPowerFlow{TR} Second Solve	40.5 ms	40.0 ms	-1.3%
ACRectangularPowerFlow{TR}(iwamoto_fallback) First Solve	163.4 ms	160.3 ms	-1.9%
ACRectangularPowerFlow{TR}(iwamoto_fallback) Second Solve	56.8 ms	40.0 ms	-29.5%

Mixed CPB

Test	Main	This Branch	Delta
ACMixedPowerFlow{NR} First Solve	5.06 s	5.478 s	+8.3%
ACMixedPowerFlow{NR} Second Solve	41.4 ms	40.9 ms	-1.1%
ACMixedPowerFlow{NR}(iwamoto) First Solve	259.3 ms	161.3 ms	-37.8%
ACMixedPowerFlow{NR}(iwamoto) Second Solve	41.5 ms	41.2 ms	-0.8%
ACMixedPowerFlow{TR} First Solve	5.263 s	5.114 s	-2.8%
ACMixedPowerFlow{TR} Second Solve	42.3 ms	42.0 ms	-0.8%
ACMixedPowerFlow{TR}(iwamoto_fallback) First Solve	163.3 ms	162.2 ms	-0.7%
ACMixedPowerFlow{TR}(iwamoto_fallback) Second Solve	61.7 ms	42.5 ms	-31.1%
ACMixedPowerFlow{LM} First Solve	7.366 s	7.033 s	-4.5%
ACMixedPowerFlow{LM} Second Solve	916.7 ms	904.6 ms	-1.3%

DC

Test	Main	This Branch	Delta
DCPowerFlow First Solve	3.826 s	3.849 s	+0.6%
DCPowerFlow Second Solve	14.3 ms	14.2 ms	-0.4%
PTDFDCPowerFlow First Solve	1.542 s	1.527 s	-0.9%
PTDFDCPowerFlow Second Solve	66.7 ms	64.1 ms	-3.9%
vPTDFDCPowerFlow First Solve	6.212 s	6.369 s	+2.5%
vPTDFDCPowerFlow Second Solve	4.196 s	4.305 s	+2.6%
DCPowerFlow First Solve	174.3 ms	193.1 ms	+10.8%
DCPowerFlow Second Solve	173.7 ms	184.1 ms	+6.0%

[jd-lara]

unconditionally zeroing it inside clear_injection_data! introduces a silent regression in the DC PF, so I'm proposing we close this in favor of handling the reset on the PowerSimulations side in Sienna-Platform/PowerSimulations.jl#1637

I tested the fix on the Argentina system too and corrected the issues found