Fix calibration

[giedriuslt]

You can set 0 or NAN as calibration constant currently, which should not be allowed, and in my case bricks the device.
Should be validated correctly.
Should fix #1502

[divadiow]

I deliberately messed up my BL0937 device again on GA then flashed to 1513_merge_7263f06183e7 - the NaNs remained through counter clears. Also, applying a power value, integer or non, gave the below. voltage and current are OK. Please sanity check me - can you replicate?

[divadiow]

if your modifications were only designed to stop it happening at all, and not fixing if already happened, then no worries.

[openshwprojects]

How does it brick a device? I was aware about the issue but I haven't seen any way it can brick stuff

[giedriuslt]

if your modifications were only designed to stop it happening at all, and not fixing if already happened, then no worries.

This fix does nothing to correct wrong data, if you have NaN in there might be hard to get out, but recalibrating and reseting energy should I believe work

[giedriuslt]

How does it brick a device? I was aware about the issue but I haven't seen any way it can brick stuff

I tested with bl602, it might have different NaN/division by zero handling. The issue was after I broke calibration, web page no longer loaded.
I tried to put it in safe mode, I could connect to device wifi, but but it still doesn't load for some reason. Although in this case it should I believe, maybe I did something wrong.
Reflashing chip which full erase (which obviously removes any config) fixed the problem

[github-actions]

🎉 This PR is included in version 1.18.21 🎉

The release is available on GitHub release

Your semantic-release bot 📦🚀