Add a general and per-module default color configuration

[Gravemind]

Add a new color_default configuration to emit a per-module default color.

If left unspecified, you get the old behavior: the non-good/degraded/bad state does not emit any color, and so you get your bar program's default color.

If color_default is specified, in general for all modules, or per-module in their respective sections, then the module will emit this color when in non-good/degraded/bad state.

Any color (color_default, color_good etc.) can be set to an empty string to disable the color, reverting to the bar program default color.

Notes on the source code changes:

• Replaced START_COLOR/END_COLOR with a variable outcolor (similar to outwalk), typed with a enum output_color_t. I think it's easier to play with the color now, and it makes it clearer it's for the entire module (as opposed to starting a color mid-sentence).
• This effectively adds a 4th state to good/bad/degraded. I thought about merging default and good state (e.g. good state by default), but some modules seem to differentiate between the default and the good state.

[Gravemind]

For what it's worth, I still use it, and, anyway, I don't mind maintaining it just for me...

I thought my last description/commit message was not great, so I reworded it to something better (I hope):

---

The significant code change was to replace the START_COLOR and END_COLOR macros by an enum and a variable outcolor used similarly to outwalk (see OUTPUT_FULL_TEXT).

A few improvements have been made from there:

1. allow any color_* config to be set to an empty string to disable it (i.e. force remove the color from the output) (also useful for 2.).

2. add a color_default config to general and to all modules/sections, so the default/normal color can be customized everywhere (color of tztime, "normal" color of memory, etc...). Unspecified by default, so same as old behavior (no color output).

3. fix per-section color emission for output_format other than i3bar.

[Gravemind]

I changed the title and description again.

FWIW I still thinks this is a good change, for users, and for the code.

I am open to suggestions. If you think this is outside the scope of i3status, I'm ok with closing this PR.