redesign bias tuning and power monitor read-out

[jordens]

• Makes bias tuning reliable on all the channels I have. Moves the algorithm to the host (close Bias tuning unreliable #66)
• Use proper asyncio sleep
• Re-implement the power monitor read-out, close ads7924 glitches #61
• Speed up the ads7924 api by packing read out of all channels.
• Speed up the channel monitoring task by packing the state machine update with the channel status read out.
• Remove the booster_channels intermediate API layer in favor of passing through the rf_channel api to closures.
• Remove a couple unnecessary constructs (RefCell, mut x: &mut y, etc)
• Reduce some code pattern duplication in mqtt_control
• Fixes an issue introduced in Fixing slope for input/reflected power measurements #74 (fixes Input power measurement wrong #59)
• Tested on a 8 channel device

NB:

• Before this, sometimes the device would reboot at random times. Presumably watchdog. My assumption is that the speed up addresses this but we should keep an eye on this. Watchdog triggering #94
• 2 Hz telemetry is probably a bit high. Maybe 0.1 Hz. telemetry rate #95
• The telemetry also stalls the mqtt api noticably. Maybe let the telemetry task handle one channel per schedule. telemetry rate #95
• Clarify that when the transforms are changed, the threshold DACs are not automatically updated. (Or change that behavior). transform setting doesn't affect thresholds #96
• It seemed to me that while the power measurements are accurate, the thresholds are wrong (too sensitive). Need to gather some data on that. check threshold accuracy #97

[ryan-summers]

Overall, I like the changes and think they're a nice direction for the code. Would you please propagate out all of the issues you have mentioned in the original PR into various issues so we can investigate in the future?