Expanding language keys and adding mixed-language speech.

[MistakeNot4892]

Description of changes

This was just supposed to be the keys stuff but I kinda spiralled.

• Languages can have keys of any length.
• Language keys must be text.
• Language keys in speech must be followed by a space.
• Using a numerical key for speech will look for that language sequentially in your language list. ie. ,1 is your first known language.
• Partial keys will use the first matching language in your language list. ,h will match ,human.
• You can use a language key in the middle of a sentence to switch language.
• Rewrote holopads to handle message passing in a better/more coherent way with less code.

Why and what will this PR improve

Mixing languages in speech is good fun for RP, and the backend changes have cleaned up, centralized a fair wodge of logic.

TODO

• Rewrite holopads to handle phrases properly. Oh god oh fuck.
• Sort out why the MC is soft resetting after speech (probably a blocking loop somewhere).
• Test basic speech (no language, one language, multiple languages)
• Test basic speech over radio.
• Test basic speech over broadcast (binary).
• Test basic speech over holopad.
• Test mixed broadcast (binary) and spoken.
• Test mixed sign/nonvocal and spoken.

Issues

• Overhead text not offset high enough.
• Animal noises are not scrambled. Type understands type, intentional.
• Sign language not transmitted over holopad.
• Cannot speak plain binary (gives 'you cannot mix non-spoken and spoken' error).

Authorship

Myself.

Changelog

TBD

[MistakeNot4892]

I need to do some more exhaustive testing before merge, but I think this is good for review now.

Stale

[MistakeNot4892]

Binary radio headset does allow you to send different languages over binary, but that is a niche enough case that I don't want to fix it for this PR.

[MistakeNot4892]

Last issue is that overhead text flickers and is out of place for 1 tick immediately after creation. I have no idea why, logic looks straightforward.

[out-of-phaze]

stopped reviewing at holopad because that's gonna take a while to get through. will look into the text over head issue in the meantime

[out-of-phaze]

a bit of code duplication but for things as hot as this it might be worth it to move the islist(mobs) check all the way out of the loop and instead have two different ones

	if(islist(mobs))
		FOR_DVIEW(var/turf/T, range, center, INVISIBILITY_MAXIMUM)
			hearturfs[T] = TRUE
			for(var/mob/M in T)
				mobs += M
		END_FOR_DVIEW
	if(else)
		FOR_DVIEW(var/turf/T, range, center, INVISIBILITY_MAXIMUM)
			hearturfs[T] = TRUE
		END_FOR_DVIEW

[MistakeNot4892]

Why? This will mean we're doing two full separate dviews.

[out-of-phaze]

i must have been sleep deprived when i wrote that, else instead of if(else). only one will run at a time, we're just hoisting the check out of the loop

[out-of-phaze]

hold on, won't this always be redundant because we added the mobs earlier when we set hearturfs? or is this for recursive contents given the get_turf? because in that case the virtual mobs system may work better for that kind of thing, idk

[out-of-phaze]

note to self: global.listening_objects_by_z could be a really good optimization here

[out-of-phaze]

note to self: moved from code/modules/mob/say.dm