Replace most uses of Map by HashMap

[reduz]

• Map is unnecessary and inefficient in almost every case.
• Replaced by the new HashMap
• Renamed Map to RBMap and RBSet for cases that still make sense (order matters) but use is discouraged.

There were very few cases where replacing by HashMap was undesired because keeping the key order was intended. I tried to keep those (as RBMap) as much as possible, but might have missed some. Review appreciated!

Bugsquad edit: Follow-up to #60881.

[reduz]

This is now ready for review.

[akien-mga]

Amended to fix (manually..) the broken formatting caused by clang-format 14 in some headers.

[akien-mga]

Did a quick sanity check as I had done for #60881. As mentioned there, it's not necessarily representative of what this PR impacts, it's just to make sure there's no huge obvious regression. The test is to start the editor on a 3D scene and then close it.

$ hyperfine -iw2 -m10 "bin/godot.linuxbsd.opt.tools.64.map_to_hashmap /home/akien/Projects/godot/projects/liblast/Game/project.godot" "bin/godot.linuxbsd.opt.tools.64.orig /home/akien/Projects/godot/projects/liblast/Game/project.godot"
Benchmark #1: bin/godot.linuxbsd.opt.tools.64.map_to_hashmap /home/akien/Projects/godot/projects/liblast/Game/project.godot
  Time (mean ± σ):     10.125 s ±  0.397 s    [User: 7.405 s, System: 0.479 s]
  Range (min … max):    9.458 s … 10.654 s    10 runs
 
Benchmark #2: bin/godot.linuxbsd.opt.tools.64.orig /home/akien/Projects/godot/projects/liblast/Game/project.godot
  Time (mean ± σ):      9.822 s ±  0.480 s    [User: 7.117 s, System: 0.491 s]
  Range (min … max):    9.016 s … 10.374 s    10 runs
 
Summary
  'bin/godot.linuxbsd.opt.tools.64.orig /home/akien/Projects/godot/projects/liblast/Game/project.godot' ran
    1.03 ± 0.06 times faster than 'bin/godot.linuxbsd.opt.tools.64.map_to_hashmap /home/akien/Projects/godot/projects/liblast/Game/project.godot'

The difference seems small enough to merge this for now and unblock other PRs.

There's likely still room for optimization and for doing proper benchmarks to see if/where there are bottlenecks. But generally speaking this PR mostly refactors the codebase to a better API so it can be merged as is, and the base HashMap / RBSet / RBMap implementations can be optimized further without changing the rest of the codebase.

[bruvzg]

Seems to run fine on macOS, and I do not see any obvious slowdowns.

Note: There are a few questionable include changes (map → rb_map while code was changed to use HashMap).

[akien-mga]

Note: There are a few questionable include changes (map → rb_map while code was changed to use HashMap).

We might want to do a cleanup pass for rb_map.h and rb_set.h by comparing the outputs of:

rg RBMap -l | sort
rg rb_map.h -l | sort

[akien-mga]

Thanks!

[lufog]

@reduz, this PR broke class_names.
ClassNameTest_G4.zip (in v4.0.alpha.custom_build [396def9] works fine)

Edit:
I just noticed that there is already a bug report for this (#61084).

[Brayan-724]

.