* fix: set correct default axes for gamepads that are not connected

`glfwGetGamepadState` will set all gamepad state variables to 0.0 if required gamepad is not connected, but `RecordAutomationEvent()` inside rcore.c expects trigger axes to be -1.0f when gamepad is not connected. Since SDL and RGFW return -1.0f in such case, this change is aligning it with them

* updated comment in rcore_desktop_glfw.c

* [rcore][android] Fix keyboard input detected as gamepad on some devices (#5387)

* [core] Add keyboard vs gamepad input test example (#5387)

Note that issue has been reported by CodeQL static analysis system

* Possibly fixed the swizzling bug

* Removed examples, and generation.

Tested on two monitors with different DPI configuration, for HigDPI enabled and not, including window resizing (with framebuffer resizing if required). Verified mouse coordinates follow the requested screen size.

Note that high-dpi awareness must be enabled by users and `CORE.Window.render` reports the scaled framebuffer size, while `CORE.Window.screen` reports the logical size.

`ToggleBorderlessWindow()` has also been reviewed to be consistent with scaling, if monitor physical display size is reported as 1920x1080 but there is a content scale of 1.5, then the borderless fullscreen window will be 1280x720, with the 1920x1080 framebuffer

This reverts commit cf0d6fc664.

This improvement is just a prove of concept, at this moment `PLATFORM_WEB` is limited in terms of software rendering by `GLFW` that only allows creating a WebGL canvas context with `glfwCreateWindow()`.

We can skip that call but then some GLFW functionality is not available (windowing, inputs). The best solution is replacing GLFW completely by a pure Emscripten implementation for `PLATFORM_WEB`.

Useful to support multiple canvases running different wasm instances in same webpage

Following raylib design, a warning log message is shown and program can continue execution.
Some early return checks have been added on most  critical functions.
[rtext] Previous implementation checking `isGpuReady` cross-module variable is not needed any more, resulting in a more decoupled code, load failure is managed at rlgl level

* Make the comments on the camera 2d fields more clear about what space each one is in.

* rlparser: update raylib_api.* by CI

---------

Co-authored-by: github-actions[bot] <41898282+github-actions[bot]@users.noreply.github.com>

- REVIEWED: Checking `NULL` input on functions getting `const char *text`, to avoid crashes
- REVIEWED: `strcpy()` usage, prioritize `strncpy()` with limited copy to buffer size
- REPLACED: `strlen()` by `TextLength()` on [rtext] module
- REVIEWED: Replaced some early returns (but keeping others, for easier code following)

* Fix modulo bias in GetRandomValue(); implement rejection sampling for uniformity

* Replace do-while with for-loop in GetRandomValue rejection sampling

target already gets assigned by the clang macro it points to, overwriting it causes it to target linux instead of android, making it check for usr directories instead of the NDK's directories

* Support 16 bit vec3 values coming from gltf

* Add support for 8 bit normals

* Use glfwSetWindowMonitor instead of Pos and Size GLFW functions

* Fix window not resetting properly when toggling out of fullscreen, formatting

Specifically "occured" -> "occurred"