refactor!: move LLM configuration from in-app settings to CLI/env vars

- Remove `api_endpoint` from Settings model and settings UI
- Add `--llm-endpoint` / `AETHERA_LLM_ENDPOINT` and `--llm-key` /
  `AETHERA_LLM_KEY` CLI flags (endpoint is required)
- Update client constructor to accept API key parameter
- Update tests and documentation to reflect new configuration approach

BREAKING CHANGE: LLM endpoint and key must now be provided via
`AETHERA_LLM_ENDPOINT` and `AETHERA_LLM_KEY` environment variables or
CLI flags instead of the Settings page.

backend/internal/client/client.go

@@ -288,8 +288,12 @@ func populateUsageTimings(msgStats *types.MessageStats, usage openai.CompletionU
 	return didChange
 }
 
-func NewClient(baseURL *url.URL) *Client {
-	oaiClient := openai.NewClient(option.WithBaseURL(baseURL.String()))
+func NewClient(baseURL *url.URL, apiKey string) *Client {
+	opts := []option.RequestOption{option.WithBaseURL(baseURL.String())}
+	if apiKey != "" {
+		opts = append(opts, option.WithAPIKey(apiKey))
+	}
+	oaiClient := openai.NewClient(opts...)
 	return &Client{oaiClient: &oaiClient}
 }