Support more texture types / GPU queries / SSBO improvements / mipmap features

[maierfelix]

A rough overview of what this adds:

• Half float texture support
• 3D texture support (binding + creation)
• Time query API to profile GPU events
• A separation of SSBO binding and updating
• Binding mip levels of image textures (used for e.g. manual mip-map generation)
• Auto mip-map generation

[aardappel]

Lets give it a better name than "Update" :)

[maierfelix]

Lets give it a better name than "Update" :)

Not sure if I can think of a better name^^

[aardappel]

Not sure if I can think of a better name^^

I can't think of a worse name :) See my changed title

[aardappel]

You have linker errors on some platforms:

Undefined symbols for architecture x86_64:
  "TimeQuery::Stop()", referenced from:
      AddGraphics(lobster::NativeRegistry&)::$_99::__invoke(lobster::Value*&, lobster::VM&, lobster::Value&) in graphics.o
  "TimeQuery::Start()", referenced from:
      AddGraphics(lobster::NativeRegistry&)::$_98::__invoke(lobster::Value*&, lobster::VM&, lobster::Value&) in graphics.o
  "vtable for TimeQuery", referenced from:
      AddGraphics(lobster::NativeRegistry&)::$_97::__invoke(lobster::Value*&, lobster::VM&) in graphics.o
  NOTE: a missing vtable usually means the first non-inline virtual member function has no definition.
ld: symbol(s) not found for architecture x86_64

[aardappel]

I guess gltimequery.cpp is not added on the mac.. I can fix that.

[aardappel]

The failures on Android/Wasm is because you use the Query functions that don't exist on those platforms, please surround their use with #ifdef PLATFORM_WINNIX (platforms that have OpenGL 4.x) and in the #else case just return some default value.

[aardappel]

Seems clang is now unhappy about:

/home/runner/work/lobster/lobster/dev/src/gltexture.cpp: In function ‘uint16_t FloatToHalfFloat(float)’:
/home/runner/work/lobster/lobster/dev/src/gltexture.cpp:37:20: error: dereferencing type-punned pointer will break strict-aliasing rules [-Werror=strict-aliasing]
   37 |     uint32_t b = (*(uint32_t*)&x) + 0x00001000;
      |                    ^~~~~~~~~~~~~

Maybe we can replace (*(uint32_t*)&x) by this kind of code:

int2float i2f;
i2f.f = x;
i2f.i

[aardappel]

Ok, lets merge this for now, we will fix issues separately