feat(kiali): Add Kiali with gevals

[aljesusg]

Kiali will run Gevals when the task filter is set to 'kiali'

[aljesusg]

First approach @Cali0707
I think it’s better to run Kiali only in certain specific cases, and not with the entire weekly set alongside the rest. WDYT?

[Cali0707]

@aljesusg I left a comment on one of your tasks, but I think it applies across most of them having looked at more of the PR so far. In general, we support llm judge on responses where we expect the LLM to be giving us information/a summary of tool findings.

In terms of when to run kiali vs. the whole set of evals I defer to @manusa - what do you think Marc?

[Cali0707]

@aljesusg do you maybe want to use the llm judge impl here to check for if the model response contains the correct info? See an example here:

kubernetes-mcp-server/evals/tasks/debug-app-logs/debug-app-logs.yaml

Lines 12 to 13 in 77b4baf

	verify:
	contains: "division by zero"

[manusa]

In terms of when to run kiali vs. the whole set of evals I defer to @manusa - what do you think Marc?

I understand that this might be beneficial to run on PRs that affect the Kiali toolset, the filter sounds like a good approach.
However, we should probably document this to avoid forgetting how the PR comments work.

Additionally, we could improve the workflow dispatch to include a drop-down of the features/toolsets to test so at some point we can dispatch an execution for a given toolset evaluation.