fix(qoder): report usage to all clients and stop inlining large attachments

- Coalesce Qoder's empty finish-in-delta frame with the later choices:[] usage
  frame so OpenAI and Claude clients receive prompt_tokens, completion_tokens
  and cache-hit tokens (the dashboard already saw them)
- Upload inlined images through /api/v2/image/upload like qodercli, and stub
  oversized non-image files instead of stuffing 30MB+ data URIs into
  agent_chat_generation
- Emit response.completed -> response.usage for chat-native upstreams so
  /v1/responses clients (Codex CLI, sub2api) no longer log 0/0/0
- Keep Claude message_delta.usage working when usage arrives without choices[0]
- Escalate to the smallest advertised Qoder context tier (200K/400K/1M) when
  the estimated prompt no longer fits max_input_tokens
- Pass apiKey for PAT connections and list hidden enable:false catalog keys
  from /v1/models

open-sse/utils/stream.js

@@ -3,6 +3,7 @@ import { FORMATS } from "../translator/formats.js";
 import { trackPendingRequest, appendRequestLog } from "@/lib/usageDb.js";
 import { extractUsage, mergeUsage, hasValidUsage, estimateUsage, logUsage, addBufferToUsage, filterUsageForFormat, COLORS } from "./usageTracking.js";
 import { parseSSELine, hasValuableContent, fixInvalidId, formatSSE } from "./streamHelpers.js";
+import { toResponsesUsage } from "../translator/concerns/usage.js";
 import { getOpenAIResponsesEventName, isOpenAIResponsesTerminalEvent, formatIncompleteOpenAIResponsesStreamFailure } from "./responsesStreamHelpers.js";
 import { dbg, isDebugEnabled } from "./debugLog.js";
 
@@ -201,7 +202,8 @@ export function createSSEStream(options = {}) {
 
               responsesTerminal = isOpenAIResponsesTerminalEvent(currentOpenAIResponsesEvent, parsed);
 
-              const isFinishChunk = parsed.choices?.[0]?.finish_reason;
+              const isFinishChunk = parsed.choices?.[0]?.finish_reason
+                || parsed.choices?.[0]?.delta?.finish_reason;
               if (isFinishChunk && !hasValidUsage(parsed.usage)) {
                 const estimated = estimateUsage(body, totalContentLength, FORMATS.OPENAI);
                 parsed.usage = filterUsageForFormat(estimated, FORMATS.OPENAI);
@@ -365,6 +367,19 @@ export function createSSEStream(options = {}) {
               item.usage = filterUsageForFormat(buffered, sourceFormat);
             }
 
+            // Responses API clients (Codex, sub2api /v1/responses): usage lives on
+            // response.completed → response.usage. Same buffer/estimate policy as above.
+            const completedResponse = item.event === "response.completed" ? item.data?.response : null;
+            if (completedResponse && typeof completedResponse === "object") {
+              if (state.usage) {
+                completedResponse.usage = toResponsesUsage(addBufferToUsage(state.usage)) ?? completedResponse.usage;
+              } else if (!completedResponse.usage && totalContentLength > 0) {
+                const estimated = estimateUsage(body, totalContentLength, FORMATS.OPENAI);
+                completedResponse.usage = toResponsesUsage(estimated);
+                state.usage = estimated;
+              }
+            }
+
             const output = formatSSE(item, sourceFormat);
             reqLogger?.appendConvertedChunk?.(output);
             controller.enqueue(sharedEncoder.encode(output));