6g3y issues

Results 5 issues of


                                            6g3y

runtime.loadJar 无法加载jar文件

![image](https://github.com/SuperMonster003/AutoJs6/assets/37627813/3952a911-ca35-4d9b-b789-72ee9d8c882c) Wrapped org.autojs.autojs.pio.UncheckedIOException: java.io.FileNotFoundException: 文件不存在或不可读: /data/user/0/org.autojs.autojs6/cache/classes/2a8cb6369908f80f87b2ee7ee9e27802 ($remote/LoadLib.js#3) Wrapped org.autojs.autojs.pio.UncheckedIOException: java.io.FileNotFoundException: 文件不存在或不可读: /data/user/0/org.autojs.autojs6/cache/classes/2a8cb6369908f80f87b2ee7ee9e27802 看描述好像是加载了class，但是在 data下面的class并没有成功加载出来

images.matToImage(mat) 无法使用

images.matToImage(mat) ![image](https://github.com/SuperMonster003/AutoJs6/assets/37627813/1df7953e-1991-4ea4-abf4-29f95e9e1687) 使用的是 importClass(org.opencv.core.Mat); 系统版本安卓13

buger repoter ：Solution to incorrect mouse position/鼠标位置不正确解决办法

问题：放缩布局不是为100%时会导致鱼饵选择失败 ![image](https://github.com/7eu7d7/genshin_auto_fish/assets/37627813/3dcf05fc-d5cf-4fb2-98cb-9801f91e74ae) 下面可以看到操作范围上只有720p，放大1.5倍后才能一致 ![image](https://github.com/7eu7d7/genshin_auto_fish/assets/37627813/324a9316-c17d-45de-91b1-72b0c37352af) 而分辨率是写死的，在获取图片的时候，直接获取长和宽的 ![image](https://github.com/7eu7d7/genshin_auto_fish/assets/37627813/d42abdcd-5c97-4446-8fb7-6f3fa2accfe3) 并且存在分辨率不一致的问题如上面的1924x1120，实际上是1920x1080 明显1120比1080多了40像素，其原因是因为窗口化带来的问题（就是上面那个白条，如下图） ![image](https://github.com/7eu7d7/genshin_auto_fish/assets/37627813/33db65c6-faba-4ee4-aedc-37186c23884a) 最终导致屏幕异常，没办法找到正确的饵料

Feature Request: PIXI.Texture.fromBuffer

### Description version : pixi.js@^8.1.3 PIXI.Texture.fromBuffer function disappearance Or allow oneself to create Uint8Array as input in sharder https://github.com/pixijs/pixijs/pull/9659 History has been added, but it has disappeared

:books: Doc Improvement

[BUG] Due to forced scaling, the output of the model's bbox is inconsistent with expectations

https://github.com/lmstudio-ai/mlx-engine/blob/main/mlx_engine/utils/image_utils.py#L18 https://lmstudio.ai/blog/lmstudio-v0.3.6 At present, pixel size scaling is hard coded in the source code, and some models require scaling to multiples of 48 to function properly. Writing dead pixels directly...