Read the extract response correctly, and fail visibly

/ocr-extract answers { extracted: {...}, raw: "..." }. The client checked
`fields` first and `extracted` second, which was right by luck — but an
extraction that returned nothing, or threw, did so silently: the button
went back to idle and the operator had no way to tell a read from a
no-read.

Now an empty or failed extraction says so under the field, and either
way the UPLOAD survives. The file is already stored and referenced; a
failed read only means the fields are not pre-filled, which someone can
recover by typing. Losing the upload because the read failed would not
be recoverable.

Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>

src/components/FileField.jsx

@@ -36,11 +36,22 @@ export default function FileField({ field, value, instanceId, activityUid, onCha
 
       if (isOcr) {
         setBusy('Reading the document…')
+        // Extraction failing must NOT lose the upload. The file is already
+        // stored and referenced; a failed read just means the fields are not
+        // pre-filled, which is recoverable by typing. Swallowing the upload
+        // because the OCR errored would not be.
+        try {
           const out = await client.ocrExtract(ref, ctx)
-        const fields = out?.fields ?? out?.extracted ?? out?.data ?? out
+          // The endpoint answers { extracted: {...}, raw: "..." }.
-        if (fields && typeof fields === 'object') {
+          const fields = out?.extracted ?? out?.fields ?? out?.data ?? null
+          if (fields && typeof fields === 'object' && Object.keys(fields).length) {
             setExtracted(fields)
             onExtract?.(fields)
+          } else {
+            setErr({ status: '', message: 'Uploaded, but nothing could be read from this document.' })
+          }
+        } catch (ox) {
+          setErr({ status: ox.status ?? '', message: 'Uploaded, but reading it failed — ' + (ox.message || 'unknown error') })
         }
       }
     } catch (ex) {