kubeai 0.9.0

Highlights

• Autoscaling now works for any engine including Ollama and FasterWhisper
• Add ability to cache models using shared filesystems (Filestore, EFS, etc)

What's Changed

• Autoscale based on KubeAI OpenTelemetry active requests metrics by @nstogner in #261
• add resourceProfiles and 405b on A100 80GB by @samos123 in #264
• Refactor e2e tests by @nstogner in #263
• Add Autoscaler State ConfigMap by @nstogner in #268
• add tpu quota to GKE install guide and use values-gke.yaml by @samos123 in #271
• update vllm images to 0.6.3 by @samos123 in #273
• Shared filesystem caching by @nstogner in #272
• add manual test of vLLM on GPU and TPU by @samos123 in #279

Full Changelog: v0.8.0...v0.9.0