fix: handle OpenAI API errors better

[akx]

I wanted to try Goose out, so I built it from source and tried configuring it with my local Ollama, only to be met with "Request failed with status: 404 Not Found". While I, knowing Ollama, could guess that meant "that model is not found", not everyone might.

This PR:

• gently cleans up error handling from OpenAI compatible endpoints by introducing a real type for them
• makes that error-handling branch handle 404 as well as 400

with the end result that the "model "qwen2.5" not found, try pulling it first" error is now surfaced.

Before

$ goose configure
Welcome to goose! Let's get you set up with a provider.
  you can rerun this command later to update your configuration

┌   goose-configure
│
◇  Which model provider should we use?
│  Ollama
│
◇  Provider Ollama requires OLLAMA_HOST, please enter a value
│  localhost
│
◇  Enter a model from that provider:
│  qwen2.5
│
◇  Request failed: Request failed with status: 404 Not Found
│
└  Failed to configure provider: init chat completion request with tool did not succeed.

  Warning: We did not save your config, inspect your credentials
   and run 'goose configure' again to ensure goose can connect

After

$ goose configure
Welcome to goose! Let's get you set up with a provider.
  you can rerun this command later to update your configuration

┌   goose-configure
│
◇  Which model provider should we use?
│  Ollama
│
◇  Provider Ollama requires OLLAMA_HOST, please enter a value
│  localhost
│
◇  Enter a model from that provider:
│  qwen2.5
│
◇  Request failed: model "qwen2.5" not found, try pulling it first (type: api_error) (status 404)
│
└  Failed to configure provider: init chat completion request with tool did not succeed.

  Warning: We did not save your config, inspect your credentials
   and run 'goose configure' again to ensure goose can connect

[akx]

Rebased post #1293 merge (which touched the same things). @yingjiehe-xyz, since you authored that one, mind reviewing this (or at least allowing CI on it to begin with)? Thanks! :)

[yingjiehe-xyz]

Thanks for the contribution! some minor comments

[akx]

@yingjiehe-xyz Rebased and took care of your suggestions :)

[akx]

Fixed rustfmt 🤦

[yingjiehe-xyz]

We have a fix for the lint: #1314 (review)

[yingjiehe-xyz]

PR #1314 (review) is merged, can you do a quick rebase?

[akx]

@yingjiehe-xyz Done deal!

[yingjiehe-xyz]

oh, sorry, just realized we had errors.rs file under the same dir, can we move these to errors.rs file instead of utils.rs?

[akx]

Absolutely. Moved! I kept the response type where it was, because it's related to response unwrapping alone and shouldn't be needed by other modules at this point :)