Fixes#25
This isolates the JKRHeap operator new/delete overloads. Every single new/delete site in the code has been replaced with a macro.
Sadly for new[] and delete[] we have to keep global operators. The global new[] just allocates into malloc() however, and delete[] goes into free() if it's not in a JKRHeap. So that's fine.
* Fix JUT_ASSERT to be a nested define
* Switch names that appear in asserts to be constants instead of defines
* Replace `0` in asserts with `NULL` or `FALSE`
* Fix fpclassify
* Fix ARRAY_SIZE
* Use G_CM3D_F_INF
* More fixes for fpclassify
* Remove FLOAT_LABEL
* Remove incorrect FLAG_ON macro
* Remove UNK_BSS macro
* Silence clangd unused header warning for PCH
* Fix missing arg to JUT_ASSERT
* Fix some MWCC version diff errors
* Compile m_Do_ext, d_demo, actor_mng
* Add VSCode task to quickly switch between versions
* Unlink magLift for debug
* Update the hash of the debug dol
The old cbea5fa... hash here was for the dol generated by the alf2dol.py script, which produces incorrect alignment.
The dol with the new hash can be obtained by using `dtk elf2dol` to convert the debug .alf file to a dol.
The DOL now builds OK.
* Fix all debug REL dtor splits
All RELs now also build OK, meaning `ninja build/ShieldD/ok` now succeeds.
* Add genMessage declarations to all HIO subclasses
* Fixing more compilation errors
* m_Do_mtx 100% on debug
Cannot be linked due to weak function name mangling?
* Improve various matches
* Fix all remaining compilation errors
* Fix new compilation errors from main
* Fix retail regression
* Link f_pc_profile_lst
PJB3005