Feed through / Downstream Lugs energy values are incorrect

[cayossarian]

The values are not properly calculated and appear to be raw synthesis of upstream lugs plus downstream values.

See discussion #230

[cayossarian]

The underlying raw Homie MQTT values are incorrect, need to open a downstream issue with SPAN.

[SPAN-API-Client-Docs](spanio/SPAN-API-Client-Docs#13)

[daking13]

After updating to 2.0.7b5, the Feed Through Power sensor improved for the better, but still seeing occassional spikes and negative values, which aren't expected. You can see the before and after here, upgrade was applied around 7:35 in the below timeline. It's still not stable or accurate as it was in 1.x with the v1 API SPAN data

[cayossarian]

Yeah I was worried about that since the panel doesn't emit the the same fidelity as the panel CT which likely gathers data at the same clock frequency when it calculates those sensors. I'm not sure what your scan interval is set to in the integration but if not already at zero and you have the CPU set it to zero to get it a bit smoother but it won't be perfect for the reasons I just mentioned.

Given that SPAN plans to fix this in the summer time frame it's difficult to justify building some Frankenstein v1/v2 beast for this one sensor. I'd recommend disabling the sensor until SPAN provides a fix as I'm likely to go back to the previous "honest" case for the sensor in v2.0.7 in the underlying package so when the defect is fixed we don't need yet another fix to unwind the beta band-aide. Alternatively you can opt to run that beta and, with a template, you can clamp the negative spikes to zero and anything over a upper threshold if you really want the sensor as best effort.