Add default-on feature to suppress duplicate reports

[gdt]

Many emitters send messages multiple times for redundancy. Often there is no protocol-level support and the 3 or 4 bursts are really one logical transmission. While e.g. receiving the temperature three times causes no problems, it's messy to send it via MQTT 3 times, and for there to be 3 update entries in a database.

This issue is a feature request for duplicate suppression to be added to the core of rtl_433, so that duplicates are not emitted on any of the output mechanisms (stdout, mqtt, influxdb, and so on). CPU time and memory requirements for the new mechanism should be reasonable compared to the amount of resources already in use. This does not impose a requirement to avoid calling data_make, and does not preclude a hash function over the data. There is no requirement to keep statistics. It is acceptable to only look back 5 seconds for candidate dups, and perhaps acceptable to look back only 1 second. Dups should probably still be eliminated if there is an intervening decode, but suppression of only adjoining dups would be a big improvement over where we are now.

There should be a flag of some kind to disable this behavior, so that people who are trying to assess reception reliability can study the full stream. But rtl_433 should default to suppression, as dups are not useful when one wants to receive the data (rather than analyze the sending system).

This is split off from #2635.

[gdt]

See #908

[gdt]

See #3018 which adds this feature to rtl_433_mqtt_relay.py, but does not resolve this issue because it does not change the base program.