Premium Request calculation for obvious failures on Github Copilot CLI side

### Describe the feature or problem you'd like to solve

When there is an obvious error on the Github Copilot CLI side, a premium request is still consumed. Premium requests are "expensive" and users would like to get value for their money.

### Proposed solution

The user feels cheated. Giving the user back the request will go a long way in building trust.

### Example prompts or workflows

I asked it to update a copilot-instruction.md file. In this process there was an error on the server side. Even asking it to check if the previous step was done, consumes another request, so this error costed me 2 premium requests:

Continuing autonomously (1 premium request)

● Response was interrupted due to a server error. Retrying...

● Response was interrupted due to a server error. Retrying...

● Response was interrupted due to a server error. Retrying...

● Response was interrupted due to a server error. Retrying...

● Response was interrupted due to a server error. Retrying...

✗ Execution failed: Error: Failed to get response from the AI model; retried 5 times (total retry wait time:
5.9623695382536726 seconds) Last error: Unknown error

I can confirm there was nothing wrong with my connectivity. I was watching something on YouTube while the agent was working...

When asking it to check if the previous step was done, I could not even change the model.

### Additional context

_No response_