[IDEA] Standardize boolean handling

[apjanke]

Oh My Zsh has a lot of variables that are treated as boolean true/false values. This includes some that the user provides to control options for plugins and so on. But we don't have a standardized representation for them.

Most of these treat 'true' as true and anything else as false. (I think.) Some treat non-empty as true (like the single-bracket [ $foo ] test). I think a couple expect '1' for true. And the tests are done differently. Some do [[ "$x" == "true" ]], some [ "$x" ], some do bash-style [[ "x$FOO" != "x" ]]. This is not a big issue; but it seems wrong that 1 is considered false in many of the option settings, and reading through boolean tests takes a little more work than it has to, IMO.

How about we standardize this? Pick a set of values that are considered true, define a function that can test those and turn it in to a $? exit status, and use that for booleans everywhere (at least user-visible ones). Say, "the strings 'true', 't', 'yes', 'on', and '1' are all considered true, maybe case-insensitively, and all other values are considered false", and a function omz_true() which tests a value, returning true ($? = 0) for the true strings and $?=1 otherwise. (List based on zsh's zstyle boolean values.)

One shortcoming is that a function call like that couldn't be used inside a [[ ... ]] test. If you need to combine a boolean test with other tests, you need to pull the omz_true outside the brackets.

# broken
if [[ omz_true "$MY_OPTION" && -n "$PROJECT_ROOT" ]] ...  

# works
if omz_true "$MY_OPTION" && [[ -n "$PROJECT_ROOT" ]] ... 

Thoughts?