Cut over public engine construction to ConfigKit sources

doc.go

@@ -1,8 +1,8 @@
 // Package promptkit provides an embeddable engine for preparing and executing
 // prompt-defined LLM workflows.
 //
-// Applications construct an [Engine] with [NewEngine], select filesystem or
-// in-memory sources and optional engine-scoped [Backend] registrations, and
+// Applications construct an [Engine] with [NewEngine], select ConfigKit
+// sources for prompts, profiles, and backends, and
 // call [Engine.InspectPrompt], [Engine.InspectProfile], [Engine.Prepare],
 // [Engine.PrepareExecution], [Engine.Run], or [Engine.RunPrepared]. Concrete
 // registries, repositories, validators, and the built-in OpenAI-compatible
@@ -17,7 +17,8 @@
 // independently. An injected [LLMClient] or [ArtifactReader] can therefore
 // still receive concurrent calls and must be safe for that use.
 //
-// NewEngine copies in-memory profiles and backend definitions. Prepare,
+// ConfigKit captures the selected configuration when NewEngine constructs an
+// Engine. Prepare,
 // PrepareExecution, and Run copy request maps, slices, pointer values, and
 // JSON-compatible extra parameters before using them. InspectPrompt and
 // InspectProfile return copied inspection values. Returned values and values
@@ -52,8 +53,8 @@
 // used by those values.
 //
 // Construction, inspection, handle, and error values, including [Config],
-// [Backend], [RunRequest], [ArtifactRef], [ExecutionTargetOverride], [Profile],
-// [OpenAICompatibleProfileConfig], [ProfileInspection],
+// [ConfigurationSources], [BackendConfig], [ProfileConfig], [RunRequest],
+// [ArtifactRef], [ExecutionTargetOverride], [ProfileInspection],
 // [PromptInputDefinition], [PromptInspection], [PreparedExecution],
 // [CapacityError], [GenerationError], [RetryExhaustedError], and
 // [ExecutionEvent], do not have stable JSON