docs: start debug / troubleshooting guide

[jgehrcke]

We can start small, but should be able to point people somewhere.

Loose ideas:

• Workload doesn't start -- now what?
• Inspect status of a ComputeDomain
• Robust kubectl-based commands for fetching 'all' current DRA driver state (we can over time work towards debug-bundle creation).
• Common problems
• How to wipe DRA driver state from a cluster (node-local state, API server state)

[jgehrcke]

How to wipe DRA driver state from a cluster (node-local state, API server state)

Came up in yesterday's GKE debug call -- while we cannot make any guarantees, we can / should provide something to 'use at your own risk'.

[jgehrcke]

Robust kubectl-based commands for fetching 'all' current DRA driver state (we can over time work towards debug-bundle creation).

I started here with documenting log collection: https://github.com/NVIDIA/k8s-dra-driver-gpu/wiki/Troubleshooting -- need to continue covering more ground, adding more detail

[shivamerla]

/cc @a-mccarthy