[esp32] Fix mgos_wifi_scan never completing if wifi in connecting state

[tripflex]

This PR
This PR adds a connecting variable to determine if wifi is in connecting state, and if so, it calls esp_wifi_disconnect before calling scan for wifi networks.

I tested this using my captive portal demo app:
https://github.com/tripflex/captive-portal-wifi-stack-demo

And it fixed the problem perfectly. Because the connect timer is still running from mgos, it will still continue connection attempt after scan is completed, moving on to next sta, etc.

---

Related to: #18
and: tripflex/captive-portal-wifi-stack#1

If wifi sta is enabled and is attempting to connect to a non-existent SSID, calling mgos_wifi_scan will never complete and mos console shows this error:

W (24763) wifi: Now is connecting, user scan invalid now!

Which if the SSID is not in range or does not exist, it will never end up connecting to anything, and scan is never actually called.

Instead of having some kind of hack in our libraries or apps, I feel as though this should be handled through the wifi lib to stop the connection attempt and actually run the scan, instead of just printing an error to console.

Related to:
espressif/esp-idf#2497

Which is related to the 3.0.x IDF versions

Looks like this is how they fixed it in wifimanager port for esp:
https://github.com/alanswx/ESPAsyncWiFiManager/pull/45/files

IDF Docs specifically reference this issue as well:
https://docs.espressif.com/projects/esp-idf/en/latest/api-guides/wifi.html#scan-when-wi-fi-is-connecting

mos-libs issue: https://github.com/cesanta/mos-libs/issues/26

#23

[tripflex]

@cpq @rojer @dimonomid can we get this merged pleaseeeeeee 👍

[bitfab]

+1 !

[tripflex]

@cpq @rojer @dimonomid

[dimonomid]

@tripflex I don't maintain Cesanta code these days, so would appreciate if you don't mention me here. Thanks.

[tripflex]

@tripflex I don't maintain Cesanta code these days, so would appreciate if you don't mention me here. Thanks.

Okay sorry, wasn't aware -- won't tag you anymore

[astout]

Thanks @tripflex, this is working for me.

[rojer]

sorry for the delay, please update the branch and address two comments

[astout]

FWIW, I spoke too soon about this working for me. I have since upgraded to 2.13.1 libs and even trying @rojer's recommended changes, I still get the "scan invalid now!" message. Would it be bad practice to just call esp_wifi_disconnect before every scan esp_wifi_scan_start in mgos_wifi_dev_start_scan? I have made this change and I am at least getting scan results now. But I am not sure if it could lead to other unknown behaviors.

[edit]
It seems to be the recommended usage now per espressif: espressif/esp-idf#2497

[tripflex]

@rojer I updated the PR per your recommendations

[rojer]

@tripflex gh says there are conflicts with the master, please rebase