Allow unsetting boolean configs

[zcsahok]

See #357.
Minor extension: on/off are also treated as true/false.

Man page update will follow when code is OK. (will have to update a lot of params...)

[zcsahok]

Updated man page. Also added some non-documented parameters.

@VictorDenisov would it be OK to rename NO_LEADING_ZEROS_SERIAL to LEADING_ZEROS_SERIAL with true default value?

[VictorDenisov]

Yeah. Totally. I guess now the configuration will look like
LEADING_ZEROS_SERIAL off
?

[zcsahok]

Yes. Any of these statements will suppress leading zeros.

LEADING_ZEROS_SERIAL=off
LEADING_ZEROS_SERIAL=NO
LEADING_ZEROS_SERIAL=False

(the argument is case-insensitive, so no does the same thing too)

[dl1jbe]

Code looks good and should catch all needed use cases. I think we should merge it in.

To drop keywords with negative logic (e.g. NO_BANDSWITCH_ARROWKEYS) it seems we need to document the default value of the boolean values in the man page.

It seems further desirable to have a command to show the state of the boolean controls in the running program.

But both may be later additions.

[zcsahok]

I'd keep legacy negative logic parameters. It would be more work to document such a change than the actual benefit. But at least for new configs we shall stick to positive logic.

Showing (and editing + reloading) configs is related to #233: one idea could be to move all globals and their initialization out of main into a data.c (like for the tests). But this is also more for the future.

[dl1jbe]

I'd keep legacy negative logic parameters. It would be more work to document such a change than the actual benefit. But at least for new configs we shall stick to positive logic.

Showing (and editing + reloading) configs is related to #233: one idea could be to move all globals and their initialization out of main into a data.c (like for the tests). But this is also more for the future.

I agree to both. They are ideas for the long run.