fix: bump llama.cpp release used in prebuilt binaries (withcatai#247)

fix: preload error on `chat` command (withcatai#245)

fix: remove CUDA binary compression for Windows (withcatai#243)

* fix: remove CUDA binary compression for Windows
* fix: improve `inspect gpu` command output

fix: bugs (withcatai#241)

* fix: avoid duplicate context shifts
* fix: `onProgress` on `ModelDownloader`
* fix: re-enable CUDA binary compression
* fix: more thorough tests before loading a binary
* fix: increase compatibility of prebuilt binaries

fix: remove CUDA binary compression for now (withcatai#238)

feat: compress CUDA prebuilt binaries (withcatai#236)

* feat: compress CUDA prebuilt binaries
* feat: automatically solve more CUDA compilation errors

feat: render markdown in the Electron example (withcatai#234)

* feat: render markdown in the Electron example
* fix: only build practical targets by default in the Electron example

fix: async gpu info getters (withcatai#232)

* fix: make GPU info getters async
* fix: Electron example build

fix: Electron example build (withcatai#230)

fix: Electron example build (withcatai#228)

* fix: Electron example build
* refactor: rename `llamaBins` to `bins`