Decompose Cilium sysdump part into fine-grained components

[marseel]

Proposal / RFE

Currently we either run all Cilium tasks or we don't run any: https://github.com/cilium/cilium-cli/blob/e04100b9f461d4e6ad58d9a3e2cccd3a3990c8d6/sysdump/sysdump.go#L1390

Ideally, we would split these tasks into groups:

• Cilium-agent tasks that are run only if we detect cilium-agent
• Cilium-operator tasks that are run only if we detect cilium-operator
• Cilium-generic tasks like collecting CRDs CNP/CCNP, Cilium configmaps etc that are always collected if any of Cilium stuff is detected
• Hubble ...
• Clustermesh ...
• Envoy
• ...

And have similar logic to https://github.com/cilium/cilium-cli/blob/e04100b9f461d4e6ad58d9a3e2cccd3a3990c8d6/sysdump/sysdump.go#L1593-L1601

This would allow us to report a bit more concise output of what was/wasn't detected at the beginning of running sysdump, something like:

Cilium components detected:

• :heavy_check_mark: Cilium Agent
• :red_square: Cilium Operator
• :heavy_check_mark: Cilium CRDs detected: [...]
• :red_square: Cilium CRDs not detected: [...]
• :heavy_check_mark: Hubble Relay
• :warning: Clustermesh - detected clustermesh-apiserver deployment, but no running pods
• ...

And then we wouldn't need to report all information for subtasks (like "Collecting ..." for things that don't exist) and only report ones that failed.