fix(modem): Fixed mode transitions between any state and UNDEF

[david-cermak]

Closes #320

[euripedesrocha]

LGTM. Left a few minor comments.

[fdkz]

Doesn't seem to fully work as intended with SIM7600.

Test case:

dce->set_mode(esp_modem::modem_mode::CMUX_MODE); // OK
physically reset modem
dce->set_mode(esp_modem::modem_mode::UNDEF); // OK
dce->set_mode(esp_modem::modem_mode::CMUX_MODE); // FAIL

esp_modem starts to send +++ and AT wrapped:
\xF9\t\xEF\x07+++5\xF9
\xF9\t\xEF\x07AT\r5\xF9

and doesn't recognize the unwrapped response by the modem:
AT\r\r\nOK\r\n

[david-cermak]

Doesn't seem to fully work as intended with SIM7600.

Test case:

dce->set_mode(esp_modem::modem_mode::CMUX_MODE); // OK physically reset modem dce->set_mode(esp_modem::modem_mode::UNDEF); // OK dce->set_mode(esp_modem::modem_mode::CMUX_MODE); // FAIL

esp_modem starts to send +++ and AT wrapped: \xF9\t\xEF\x07+++5\xF9 \xF9\t\xEF\x07AT\r5\xF9

and doesn't recognize the unwrapped response by the modem: AT\r\r\nOK\r\n

That's not how the UNDEF mode is supposed to work. It just a meta-state indicating that all transitions are allowed, not that they'd necessarily succeed (nor it is a synonym to INIT or RESET state).

What you describe in your test case is more like a situation suitable for mode detection (another planned feature), in other words: We don't know what state the modem is in, so we check how it responds to all protocols and set the mode accordingly.