docs: fix nvidia-container-runtime usage example

[iacker]

This updates the nvidia-container-runtime usage example so it follows the runtime path described earlier in the README: the NVIDIA runtime modifies the OCI spec when a create command is detected, then the container can be started.

The old example used run, which skips the modifier path and leaves the generated config.json unchanged, as reported in #781.

Fixes #781

Validation:

• Confirmed the old nvidia-container-runtime run nvidia_smi example is no longer present.
• Confirmed the replacement example uses create, start, and cleanup with delete.

AI assistance disclosure:

• I used AI assistance to identify the stale example and prepare this small documentation update.

[copy-pr-bot]

This pull request requires additional validation before any workflows can run on NVIDIA's runners.

Pull request vetters can view their responsibilities here.

Contributors can view more details about this message here.

[henry118]

LGTM

[cdesiniotis]

I tried out the new procedure, and hit the below error:

$ sudo nvidia-container-runtime create nvidia_smi
ERRO[0000] runc create failed: cannot allocate tty if runc will detach without setting console socket

I have modified the commit in this PR so that the procedure is functional.

@henry118 ptal

[henry118]

LGTM. Thanks for pointing out the TTY issue.

[cdesiniotis]

Thank you @iacker for the contribution.