Background HR

[mark9064]

Largely based on #1718

Changes:

• 4 state machine
• Heart rate display avoids stale values being shown for a long time in some scenarios
  • does not cover all scenarios, but hopefully good enough and as close to the current behaviour as possible while still making sense for background measurement
• Constant frequency HR measurements that take into account successful screen on measurements (and also long periods of unsuccessful screen on measurements)
• Refactored a few annoying bits of the PPG code like lastBpm and the bodge to get correct not enough data messages
• Removed the 15s mode. The sensor already takes ~6s to get a measurement in ideal conditions so it makes sense to use continuous - the sensor on time will be very similar and proper history averaging with continuous means measurement accuracy is better
• Misc code bits such as using std::optional more where sensible. I haven't used it in the settings struct as the layout is compiler dependent AFAIK

Accuracy improvements (second commit)

• Sampling is constant frequency instead of delay based
• Raised the priority of the HR task to 1 so that it does not get out-competed by DisplayApp

This is still more complex than I'd like it to be