Move texture2d mapping to renderer gl

docs/modernization/build-inventory.md

@@ -133,7 +133,9 @@ Known local toolchain state:
   invalid channel counts rejected by `Texture2D::create(Image)`, RGBA8/RGBA32F
   readback formats, byte-count math, and PBO pixel-buffer target/usage/access
   mapping used by `RTT` and `PBO` readbacks, and framebuffer status naming
-  used by `RTT` diagnostics. It also owns and
+  used by `RTT` and `Texture2D` diagnostics. It also owns the 2D texture target,
+  framebuffer setup, readback format, mipmap target, and update component-type
+  tokens used by `Texture2D`. It also owns and
   validates framebuffer blit color mask and linear/nearest filters used by
   `RTT::resize` and `RTT::copy`, plus the default linear clamp-to-edge
   render-target texture parameters, texture/renderbuffer targets, depth format,

docs/modernization/roadmap.md

@@ -391,7 +391,9 @@ float texture support. It also owns the OpenGL texture upload-type mapping used
 by legacy `Texture2D` and `RTT` creation, RGBA pixel-format mapping used by
 `RTT` texture allocation, plus image channel-count to texture
 format mapping for `Texture2D` image uploads and framebuffer status naming for
-`RTT` diagnostics. RGBA8/RGBA32F readback formats, byte-count math, and PBO
+`RTT` and `Texture2D` diagnostics. `Texture2D` 2D texture binding, upload,
+mipmap generation, framebuffer readback setup, and update component-type tokens
+now delegate to `pp_renderer_gl`. RGBA8/RGBA32F readback formats, byte-count math, and PBO
 pixel-buffer target/usage/access tokens used by `RTT` and `PBO` readbacks now
 live in `pp_renderer_gl`. The framebuffer blit color mask and linear/nearest
 filter tokens used by `RTT::resize` and `RTT::copy`, plus the default