BROKER_REAL_TIME=1 causes crashes in the INFLUXDB module

[mcspr]

The problem is with POW R2, it crashes few minutes after reboot when BROKER_REAL_TIME is set to 1. In web debug i see INFLUXDB messages in every 5-6 seconds.

Originally posted by @MihataBG in #1968 (comment)

---

90% certain this is related to the issue with ESPAsyncTCP, fixed by the me-no-dev/ESPAsyncTCP@75c2513

I don't really remember why exactly I skipped the update, besides the ASYNC SSL build problem. Need to reassess this again.
Or, do things like #1806 and just use AsyncClient directly.

[MihataBG]

Also have a problem with INFLUXDB on "magichome-led-controller". I tried with BROKER_REAL_TIME=0 but module restart itself when do INFLUXDB message.

[kordian-kowalski]

Same here, enabling InfluxDB will crash a Wemos D1 mini after sending a couple of measurements, and I'm also seeing the measurements being sent more like every 6 seconds rather than 10 that I set up.

[mcspr]

Right, previous version only sent influx message each sensor report, while BROKER_REALTIME=1 now makes it every reading.
Considering I already broke broker apart, I wonder if it is also viable to just remove REALTIME flag altogether and just have SensorReportBroker and SensorReadingBroker and make influx subscribe to report one.

Problem with SyncClient that I mentioned previously is quite random and depends on how well network behaves, sending things too frequently shows it is thoroughly broken. See me-no-dev/ESPAsyncTCP#115 explanation, but we need to throw out old SSL stuff first.
upd: I wish Core's HTTP client had any request argument stuff :/ But leaning to adapting Thingspeak logic / #1909

[mcspr]

#2061 reverts to the old behaviour and sends data every report, please check again with the current dev build

There are also two additional changes:

• string measurements are quoted
• http request now sends multiple measurements instead of one at a time.