Fixed Point Math

[SBird1337]

Provides inlined fixed point math operations for Q_4_12 FP format.

Description

Introduces new type ufp_4_12_t for unsigned fixed point numbers in specified format, as well as inlined version of add, sub and multiply.

Refactors battle damage modifier calculations to use the new functions in order to improve readability and performance. (See Discord conversation)

This PR does not alter any of the other usages of FP math.

Discord contact info

Karathan

[mrgriffin]

Looks good to me, so some minor naming nitpicks and that's it:

• It would be nice if UQ_4_12 and ufp_4_12_{t,add,sub,multiply} used the same names, but that'd probably mean uq_4_12_{...} rather than renaming UQ_4_12 (so that we don't break uses elsewhere) and idk how you feel about that.
• add, sub, multiply breaks the pattern for me, what's your feeling about mul instead? Or I suppose subtract 🤔

Regardless, I'm very pleased with this change! It's great that the typedefs are 32-bit :)

[SBird1337]

Looks good to me, so some minor naming nitpicks and that's it:

• It would be nice if UQ_4_12 and ufp_4_12_{t,add,sub,multiply} used the same names, but that'd probably mean uq_4_12_{...} rather than renaming UQ_4_12 (so that we don't break uses elsewhere) and idk how you feel about that.
• add, sub, multiply breaks the pattern for me, what's your feeling about mul instead? Or I suppose subtract 🤔

Regardless, I'm very pleased with this change! It's great that the typedefs are 32-bit :)

I think Q_ is fine, even though I think people are usually more familiar with the concept of fixed point and thats what we are calling internally. For the sake of not having to refactor everything else we should probably stick with Q / UQ.

How do we feel about the function names in general. I kind of broke convention with CamelCase which is what we usually do. Same about the types. We do not usually add a _t suffix.