Memory improvements + Type analysis

[BobVarioa]

Summary:

• adds an allocString method to the allocator, currently only for the static allocator
• changes String.prototype.concat to be a builtin method, instead of a function in codegen
• removed all Porffor.bs/s calls
• adds Porffor.allocatePage<type> and Porffor.allocateBytes<type> as simple memory management utilities
• rewrites some methods to allocate bytes instead of pages
• reenables js type inference (this allows String.prototype.concat to function in more cases, along with many other methods)
• adds Porffor.allocateNamedPage<type>(key) to create a pointer to a single keyed place in memory

Current issues: (box is checked if it is completed)

• Precompiled strings are not yet included at runtime
• All strings should be initialized at the beginning of a program, and not at various locations throughout
• makeStringBuffer should be removed, as it has been superceded by the Porffor.allocate methods
• Improve compiler performance for precompiled strings, it's currently at unacceptable levels
• Investigate the -0.09 percent test262 drop
  • Most of these seem to be uncovered issues? i.e. there is a lot of new TypedArray failures, but that's mostly because we don't have ArrayBuffer

Future extensions:

• Avoid Porffor.allocatePage whenever possible, choosing Porffor.allocateBytes instead
• Convert the current array implementation to a linked list so that it can actually have the benefits of allocateBytes (set as well)
• Actual malloc and free, to be any useful requires a GC so likely far future

[CanadaHonk]

Thanks! Commented

[BobVarioa]

I still think this PR is not quite ready to be merged, but I think it's at a stage where this is no longer a draft

[CanadaHonk]

We're going to slowly incrementally add the improvements from this into main instead of just merging one chunky PR :)