feat: add SDXL support (leejet#117)

* add SDXL support

* fix the issue with generating large images

feat: implement ESRGAN upscaler + Metal Backend (leejet#104)

* add esrgan upscaler

* add sd_tiling

* support metal backend

* add clip_skip

---------

Co-authored-by: leejet <[email protected]>

feat: implement the complete bpe function (leejet#119)

* implement the complete bpe function
---------

Co-authored-by: leejet <[email protected]>

fix: enhance the tokenizer's handing of Unicode (leejet#120)

fix: add support for int32_t on other compilers (leejet#114)

fix: remove dangling pointer to work_output in CLIPTextModel (leejet#111

)

feat: add SD-Turbo support

chore: make code cleaner

fix: generate image correctly in img2img mode

feat: add TAESD implementation - faster autoencoder (leejet#88)

* add taesd implementation

* taesd gpu offloading

* show seed when generating image with -s -1

* less restrictive with larger images

* cuda: im2col speedup x2

* cuda: group norm speedup x90

* quantized models now works in cuda :)

* fix cal mem size

---------

Co-authored-by: leejet <[email protected]>