Add RzIL for floating-point x86 instructions

[DMaroo]

Your checklist for this pull request

• I've read the guidelines for contributing to this repository
• I made sure to follow the project's coding style
• I've documented or updated the documentation of every function and struct this PR changes. If not so I've explained why.
• I've added tests that prove my fix is effective or that my feature works (if possible)
• I've updated the rizin book with the relevant information (if needed)

Detailed description

Add RzIL implementation for x86 floating point instructions. This is a WIP currently.

All the unrelated commits will go away once #3864 gets merged. Done.

Test plan

Added tests in db/asm/x86_64. Pretty much all the instructions are covered. May add more tests depending on the code coverage.

Tracetesting has been put on hold for now (see this comment), but will add the pending tracetesting in the tracker issue.

Closing issues

Closes #3861.

[XVilka]

I also recommend taking some binary and targeting FP instructions located in it. Either one already in rizin-testbins, or take/compile a new one. Then, it can be tested as early as a subset of the instructions used is implemented.

[DMaroo]

I'm planning to write a testing 2D matrix multiplication code for floating point numbers to test the IL. I am not done with the arithmetic FP operations, so I need to lift them first before adding such a test. Maybe some programs from here can also be used: https://benchmarksgame-team.pages.debian.net/benchmarksgame/index.html.

[DMaroo]

50+ x86 floating point instructions have been lifted to RzIL! This is a significant majority of the x86 floating point instructions (discount SSE, MMX, AVX and other such ISA extensions). Most of the remaining instructions are either too complicated and rare to bother implementing or use transcendental operations unsupported by RzIL or are lower-level instructions which would not have any equivalent reasonable representation in RzIL. I believe the lifted instructions are more than enough to allow for a very satisfactory usage experience.

I need to now add asm tests for all these instructions and then perform tracetesting. I am not adding any more instruction liftings until the testing of the existing liftings is over.

[DMaroo]

ASM tests pass with the existing x86 FPU instructions in db/asm. Will probably add some more instructions in the test file before moving forward with tracetesting.

[XVilka]

Even though there is no emulateme test, I propose merging this before the release, just mention that this support is experimental in the release notes, and will be improved in the next one. What do you think?

[DMaroo]

I agree about getting this merged. It is incorrect for some floating-point operations (due to the 80-bit floating-point shenanigans), but I'm confident that the rest of the instructions function as expected.