Fast models on databricks have 0 retries.

[katzdave]

No description provided.

[Copilot]

Pull request overview

This PR ensures that Databricks “fast” models (and embeddings) use a zero‑retry policy so they fail quickly instead of being retried with the standard backoff configuration.

Changes:

• Extends the ProviderRetry trait with a configurable with_retry_config helper and reimplements with_retry in terms of it.
• Adds a separate fast_retry_config to DatabricksProvider, hardcoded to 0 retries, wired through both environment- and parameter-based constructors.
• Updates Databricks completion and embedding calls to use fast_retry_config when invoking the fast model (and always for embeddings), while preserving existing retry behavior for non-fast models and streaming.

Reviewed changes

Copilot reviewed 2 out of 2 changed files in this pull request and generated no comments.

File	Description
crates/goose/src/providers/retry.rs	Adds a configurable with_retry_config method to the ProviderRetry trait and makes with_retry delegate to it, enabling per-call retry configuration.
crates/goose/src/providers/databricks.rs	Introduces a dedicated 0-retry fast_retry_config for Databricks, wires it into provider construction, and applies it to fast-model completions and embeddings while leaving standard retry behavior unchanged elsewhere.