installing CML

### v1

```mermaid
graph TD
start[How to Install CML] --> Dq{Using your own Docker image?}
Dq-->|Yes| Gq{Using GitHub?}
Gq-->|Yes| SC["iterative/setup-cml@v1"]
Gq-->|No| MI[cml.dev/doc/install]
Dq-->|No| Nq{Using a GPU?}
Nq -->|Yes| DG[iterativeai/cml:0-dvc2-base1-gpu]
Nq -->|No| D[iterativeai/cml:0-dvc2-base1]

click SC "https://github.com/iterative/setup-cml" _blank
click MI "https://cml.dev/doc/install" _blank
click D "https://cml.dev/doc/self-hosted-runners#docker-images" _blank
click DG "https://cml.dev/doc/self-hosted-runners#docker-images" _blank
```

### [wip] helio v1

```mermaid
graph TD
start[Install CML] --> Gq{GitHub Actions?}
Gq-->|Yes| SC["uses: iterative/setup-cml@v1"]
Gq-->|No| Nq{Have npm?}
Nq-->|Yes| Mn["npm install -g @dvcorg/cml"]
Nq-->|No| Mc["curl https://cml.sh/ | sh"]
```

### [wip] helio v2

```mermaid
graph TD
start[How to use CML with containers] --> Sq{Just trying out CML?}
Sq-->|Yes| SC["use iterativeai/cml container images"]
Sq-->|No| Nq{Is dependency install network-bound?}
Nq-->|No| Mn["Build your own container image"]
Nq-->|Yes| Mc["Use a standard base image"]
Mn --> Gq{Do you need GPU support?}
Mc --> Gq
Gq-->|Yes| Mv["e.g. nvidia/cuda:11.2.1-cudnn8-runtime-ubuntu20.04"]
Gq-->|No| Mu["e.g. ubuntu"]
```

- [ ] cross-reference (e.g. https://github.com/iterative/cml.dev/pull/423/files#r1046467075)
- related: #381