[LoRaWAN] Major rework

[StevenCellist]

@jgromes best of luck... :)
Culmination of a whole summer of development.

Totally non-exhaustive list of changes in somewhat random order of importance:

• Implement LoRaWAN v1.0.4 (for real), alongside v1.1
• Implement Regional Parameters v1.0.4 (v1.1 now only available as fallback)
• Rework MAC command processing ...
  • ... repeated MAC commands are now implemented (until downlink is received)
  • ... not tied anymore to arbitrary payload length and queue length (sacrificing some ease of use, but some optional MAC commands completely invalidated the previous system with a length of up to 26 bytes)
  • ... ADR is now pre- and post-processed into a complete channel mask and processes & responds as required for both v1.0.4 and v1.1 on dynamic and fixed bands (this was honestly a hell)
  • ... configurations are now only applied if the whole MAC command succeeded
  • ... and some misc bugs
• Equally use all available channels by keeping track of used channels
• Support frame retransmission as part of ADR
• Implement TR013 = CSMA = LBT (for real)
• Remove uplink() and downlink() as this separation is not possible with ADR and LBT
• Don't receive downlinks that are (much) longer than allowed
• Prevent stack crashes in case of slight user error (resolves RADIOLIB_ERR_UPLINK_UNAVAILABLE (-1108) on TIG stack sensor in the field (LoraWan/TTN/InfluxDB/Grafana) #1175)
• Implement actual Rx1-datarate lookup table as RP 1.0.4 doesn't allow being smart
• Fix maximum payload calculation
• Add option to schedule Join/Uplink at specific time
• Add virtual tag to some functions to allow a derived class
• Slights hints of FSK and LR-FHSS modem, requires PhysicalLayer::setModem() to be implemented before completely working

Additional changes:

• Return Rx window number (0 for no downlink, > 0 for downlink) instead of NO_DOWNLINK or ERR_NONE - this is a 'breaking' change.
• Also use getMaxPayloadLen() for non-dwell-time bands for ease of use under ADR.
• Update some return codes to reflect error state.

[HeadBoffin]

Add option to schedule Join/Uplink at specific time

Is this required by some corner case on the spec - because it's the simplest way to DDOS a gateway - have all your nodes set to uplink at the top of the hour ...

It also contradicts TR007 with jitter - randomisation of uplink time - I usually use ±60 seconds for deployment.

Making the function inconspicuous is fine.

[StevenCellist]

It's not in TS001, it is in another TS.. but I can surely make it private.

[StevenCellist]

Edited original post to reflect additional change.

[StevenCellist]

@jgromes all done (plus minor edit of original post)

[jgromes]

Merged - great job @StevenCellist! I ran the full CI manually and it still fails in couple of places, but that's unrelated to this PR, so I will fix it later.

[StevenCellist]

Woohoo! 🎉