Function resolve_model_limits 

pub fn resolve_model_limits<'a>(
    inference: &InferenceConfig,
    context: Option<&ContextConfig>,
    driver_manifest: impl FnMut(&str) -> DriverManifest<'a>,
) -> Result<Vec<ChoiceModelLimits>, ModelLimitsError>

Resolves the output cap, context window, input ceiling and input budget of every driver choice inference declares, the primary first and the alternates in manifest order. Empty under transport: process.

driver_manifest takes a choice’s driver as inference names it (its tool name) to that driver’s label and bundled manifest. Each driver’s table is parsed once.