feat: streamv2v (#167)

* feat: streamv2v init

* fix: dynamic streamv2v

* fix: controlnet

* fix: cache res hardcode

* fix: run without streamv2v

* fix: kvo_cache_strcuture

* fix: dynamic shape hardcode

* fix: cleanup v2v features

* fix: cache interval

* fix: param updater

* fix: cleanup

* fix: diffusers package

* fix: reqs

* Fix dependency issues

* update diffusers version

---------

Co-authored-by: Victor Elias <victor@livepeer.org>

fix: handle shard models in offline mode (#160)

upgrade: upgrades safety_checker model (#122)

* upgrade: upgrades safety_checker model

* fix: safety_checker threshold level bug

* fix: update stream params

* fix: revert nsfw back

* Update src/streamdiffusion/acceleration/tensorrt/runtime_engines/unet_engine.py

---------

Co-authored-by: Victor Elias <victor@livepeer.org>

wrapper: Fix controlnet and safety_checker loading (#120)

* wrapper: Always load the safety checker if available

* wrapper: Respect use_controlnet field

We've been completely ignoring it if we don't send any controlnets on load.

The module shoud be loaded in any way. Controlnets may be added or not.

* Allow updating safety checker threshold dynamically

* Fix controlnet_config processing

Merge pull request #75 from livepeer/vb/feat/dynamic_t_index

feat: dynamic t_index support

fix: controlnet loading (#74)

* fix: controlnet loading

* fix: remove print

setup: Pin cuda-python version before v13

Set to the exact version we know works on the latest ai-runner
image:
```
$ docker run --gpus all livepeer/ai-runner:live-app-streamdiffusion conda run -n comfystream pip freeze | grep cuda-py    (base)
cuda-python==12.9.0
```

fix normalized linear interpolation for seeds

Multi-Resolution (Pipeline Reload) + WebUI Upgrades (#15)

* Added api-only mode to server, update svelte vite and frontend packages, built multiresolution support with tensorrt, built resolution picker.

* Added realtime resolution updates.

* Small fix.

* Controlnet work.

* ControlNet testing.

* Rebuilt resolution swapping by restarting pipeline, re-added prompt input to UI for testing.

* Fix missing file.

* Fix to config updating the resolution.

* fix input profiles

* restore sdxl functionality

* remove legacy prompt box

* fix prompt blending

* unified blending control and fixes

* Set dynamic size from 384-1024 with opt at 704.

* updater: Remove "dynamic" resolution update logic

* Remove debug logs

* Remove more debug logs...

---------

Co-authored-by: RyanOnTheInside <7623207+ryanontheinside@users.noreply.github.com>
Co-authored-by: Victor Elias <victorgelias@gmail.com>
Co-authored-by: Victor Elias <victor@livepeer.org>

Merge pull request #14 from livepeer/ryan/feat/mediapipe-pose-improve…

…ments

Ryan/feat/mediapipe pose improvements