ESP32: support dynamic freq scaling and wifi power save

[tve]

This PR allows the power consumption of the esp32 to be tweaked. It enables:

• using dynamic frequency scaling allowing FreeRTOS to reduce the clock frequency when the processor is idle
• using listen interval on the WiFi STA interface to reduce the amount of time the radio is powered on
• using listen interval to keep the radio on permanently to reduce network latency

The punchline is that one can halve the power consumption for more battery life or double it for lower network latency.

These features are documented, see the PR files. I have not been able to preview the docs, however, so I assume there are issues. I need to figure out how to do that...

One caveat I'm just realizing is that I haven't tested with ESP-IDF v3....

[tve]

AN issue I just discovered: the following loop in uselect busy-waits: https://github.com/micropython/micropython/blob/master/extmod/moduselect.c#L250-L257. I'll throw a ulTaskNotifyTake(pdFALSE, 1); in there to see what happens...

Edit: also https://github.com/micropython/micropython/blob/master/extmod/moduselect.c#L176

[tve]

Well, this last commit fixed asyncio. But now I better profess that I have no friggin' clue about what I'm doing, LOL :-)

[tve]

rebased on master and squashed commits

[moefear85]

I'd like to mention, that as per the esp-idf documentation, it is possible to keep some peripherals, including UART, functioning properly during auto-lightsleep (specifically frequency scaling), by using "REF_TICK" as a tick source. That would make this feature far more usable and desirable.

[dpgeorge]

I think the WLAN.connect.(..., listen_interval) addition is great, definitely useful and even necessary for some applications.

The dynamic frequency scaling is also good but I think it's not yet clear how to make it work properly. And the changes here (like changing the UART clock source) seem like they could have unintended affects on existing behaviour. Also newer version of the IDF (since v4.1) have changed how this is configured.

So I suggest to split out the addition of listen_interval and merge just that feature.

[EternityForest]

What if auto lightsleep was something you manually enabled and disabled? Lots of applications don't need to do anything besides listen for commands, most of the time, you could just manually come out of that state when needed.

[projectgus]

This is an automated heads-up that we've just merged a Pull Request
that removes the STATIC macro from MicroPython's C API.

See #13763

A search suggests this PR might apply the STATIC macro to some C code. If it
does, then next time you rebase the PR (or merge from master) then you should
please replace all the STATIC keywords with static.

Although this is an automated message, feel free to @-reply to me directly if
you have any questions about this.