utility: Implement ParemeterBase::createByTypeName

[german77]

Supersedes #22 by implementing the actual function. This has the same effect as previous PR but keeps all the previous matches intact and adds 77 new matches. This also removes a hack with ParameterCurve that for some reason was added to the header.

The function itself is not matching but really close. Most differences are in ParameterCurve and ParameterBuffer initialization. The inlining here is a killer, ghidra doesn't want to assign the correct types, the branching is chaos and reading asm is not fun in a function with this size.

I will keep working on it but I don't expect any significant improvements in a few weeks.

https://decomp.me/scratch/N8MEk

---

This change is 

[leoetlino]

The inner cast can just be a static_cast, but yeah the const_cast looks a bit dodgy. Is it possible that mValue is const qualified?

[german77]

AL has almost a copy paste implementation of ParameterBase. So far mValue is only const for cstrings types.
https://github.com/MonsterDruide1/OdysseyDecomp/blob/master/lib/al/Library/Yaml/ParameterBase.h#L191

[leoetlino]

are you sure the hpp doesn't exist? I could be misremembering but I think it might have been done this way because of some asserts in Labo. though I guess it doesn't really matter

[german77]

I don't have any file info. All I can see are inlined calls and it looked really weird to have the cpp directly included in the header

[german77]

I reverted the ParameterCurve changes. It makes no difference for this function, if an assert says the file is this way I have no problem with that.

[leoetlino]

yeah so I've just checked Labo and it looks like the asserts for ParameterCurve are in a file called aglParameterCurve.hpp.

note that it's .hpp, not .cpp, and we know Nintendo does this kind of thing as seadSafeString.h/hpp/cpp is a thing too

[german77]

The function has been matched! Please run the workflow.
There is a hack on ParameterCurve reset. Apparently it doesn't inline without doing that.