PAL-1217 heatlhbench original

[pratacosmin]

this can be run like so
define in config the model_deployments.yaml
and run

 helm-run --run-entries "health_bench:model=jsl/medical-visual-lm-30b,jury_config_path=<PATH>/judges.yaml" --max-eval-instances 4 --suite my-suite --local-path ./config

helm-summarize --suite my-suite -o ./benchmark_output --schema-path  <PATH>/schema_medhelm.yaml

 helm-server --suite my-suite -o ./benchmark_output --port 8000

[chakravarthik27]

LGTM