STM: monotonic time even when RTC is changed

[dhalbert]

• Fixes Setting the rtc on STM32 affects the value of time.monotonic() #7808

Test:

>> import time,rtc
>>> r = rtc.RTC()
>>> print(time.monotonic()); r.datetime = time.struct_time((2023,1,1,0,0,0,0,0,0)); print(time.monotonic())
1329.49
1329.49
>>> print(time.monotonic()); r.datetime = time.struct_time((2022,1,1,0,0,0,0,0,0)); print(time.monotonic())
1339.38
1339.38
>>> print(time.monotonic_ns()); r.datetime = time.struct_time((2022,1,1,0,0,0,0,0,0)); print(time.monotonic_ns())
1353792083741
1353791992187
>>> print(time.monotonic_ns()); r.datetime = time.struct_time((2024,1,1,0,0,0,0,0,0)); print(time.monotonic_ns())
1362228332530
1362227539062
>>> 

I also tested after ctrl-D: time.monotonic() is not reset to zero, which I think is OK.

The change looks bigger than it is: I had to reorder some things to avoid a forward declaration.

[jepler]

makes sense, didn't test.

[Neradoc]

This fixes #7808 from my tests on a Feather STM32F405 Express.
Thanks !

[dhalbert]

I need to look at this again. I didn't notice time is going backwards slightly here:

>>> print(time.monotonic_ns()); r.datetime = time.struct_time((2022,1,1,0,0,0,0,0,0)); print(time.monotonic_ns())
1353792083741
1353791992187

[dhalbert]

monotonic_ns() now always goes forward. subticks (32 per tick) were not being taken into account.

>>> print(time.monotonic_ns()); r.datetime = time.struct_time((2022,1,1,0,0,0,0,0,0)); print(time.monotonic_ns())
35930114752
35930236824

I am not so fond of the STM32 RTC peripheral.

[tannewt]

Thank you!