Elliot Gunton

Need to play around with some examples, but initial thought is figuring out that when the user has arguments like ```py arguments={"dataset_configs": [x_dataset, y_dataset]} ``` That get converted to: ```yaml...

Thanks for the comment @Joibel, I didn't actually see that in the docs! I originally thought this worked for templates within the workflow i.e. not a `templateRef`, but upon double...

I had an idea - but I'd be wary of doing something not obviously backwards-compatible or that blocks us in future - which would be to import all the models...

For Hera's lifecycle hooks, they currently require a full model `LifecycleHook` object, as [there is no special Hera class](https://github.com/argoproj-labs/hera/blob/4bb24fbbde94e447adb7ba579ac07e408da536dc/src/hera/workflows/workflow.py#L211) for these yet. [See example here](https://hera.readthedocs.io/en/stable/examples/workflows/upstream/exit_handler_with_artifacts/). I can take a look...

Discussed in maintainers meet up - we can just edit pydantic config https://docs.pydantic.dev/1.10/usage/model_config/#options `extra` > whether to ignore, allow, or forbid extra attributes during model initialization. Accepts the string values...

Fix was reverted in main due to impacting the `WorkflowsService`, reopening. We now have "on_cluster" tests using the WorkflowsService so at least that won't happen again, but we need to...

Do you think we can call this fixed if we add some README/[quick start guide](https://hera.readthedocs.io/en/stable/walk-through/quick-start/) instructions to use `make install-k3d install-argo run-argo` using the repo itself in e.g. github codespaces?...

Also see #961

On the Hera dev side, the new RunnerIO code in `runner.py` has made it pretty difficult to understand/navigate. Needs some tidying up and refactoring. cc @samj1912 @flaviuvadan

Script annotations have been superseded by Pydantic IO which makes functions much less verbose - ~is there any reason to keep raw `Annotated` function params around or should we push...