Add new LegacyDialog entity

[Hammie]

Adds a LegacyDialog entity that implements the old Seadogs 2 dialog system for use in New Horizons.

Does not yet include pagination or mood settings for the head animation.

Also includes #383

[q4a]

@Hammie sorry, but clang is more strict. If you create td::vector<LinkEntry> formattedLinks_; then you need to emplace_back only LinkEntry. I fixed problem, builded on Linux and tested a bit.
Works fine, thanks!
2ff2085

             for (const auto &text : link_texts)
             {
-                formattedLinks_.emplace_back(text, static_cast<int32_t>(i));
+                formattedLinks_.emplace_back(LinkEntry{text, static_cast<int32_t>(i)});
             }

[Hammie]

@q4a Thank you very much!

Nothing quite like waking up to see problems already solved 👍

[espkk]

It looks good!
I see you removed some utf8-related code from the original Dialog though. Did you mean to drop its support or it's still supported somehow?

PS
@q4a, iirc from our discussion clang is not more strict, it just doesn't support this C++20 feature yet 😅

[Hammie]

I see you removed some utf8-related code from the original Dialog though. Did you mean to drop its support or it's still supported somehow?

@espkk I assume you are talking about the modified AddToStringArrayLimitedByWidth? I'm confident the new code still works with utf-8, but I will try and find a more conclusive way to make sure.

[Hammie]

@espkk I've tried creating a unit test to make sure it still works with non-ascii characters, but the presence of the renderer makes that almost impossible.

The theory as to why it should still work is that neither ' ' or '\n' could ever by confused for other characters in utf-8 - as is the case for all ascii characters <= code point 127 - because multi-byte code points always have the highest bit active.

Also did a quick test with some Cyrillic text, no idea what it says but it looks like it is rendering correctly:

[espkk]

@espkk I've tried creating a unit test to make sure it still works with non-ascii characters, but the presence of the renderer makes that almost impossible.

The theory as to why it should still work is that neither ' ' or '\n' could ever by confused for other characters in utf-8 - as is the case for all ascii characters <= code point 127 - because multi-byte code points always have the highest bit active.

Also did a quick test with some Cyrillic text, no idea what it says but it looks like it is rendering correctly

Sounds good! Thanks

But this needs a rebase now 😢

[espkk]

only 1 minor note - please move new header ito src if it's not public
include is for public headers which are getting exported