Improve PrinterFarm reliability and batch uploads

[ashimaryal25]

What changed

• Recover print percentage from layer telemetry when Creality reports a stale zero value.
• Remove duplicated percentages from printer job labels.
• Retry slow discovery probes, use a realistic timeout, merge results without duplicates, and prevent background polling from racing discovery.
• Replace stale online snapshots when a printer becomes unreachable.
• Require operators to confirm an idle bed is clear after startup or discovery, without hiding active prints.
• Allow multiple G-code files to be selected and queued in one action.
• Keep the first-run empty state focused on network discovery.

Root causes

Creality printers can report fragmented or delayed telemetry, and their status socket tolerates few simultaneous connections. PrinterFarm trusted the percentage field even when layer counters advanced, scanned each address only once with a timeout shorter than observed printer response times, allowed discovery and polling to contend for the same socket, and preserved stale online telemetry after complete network failures. The upload picker also consumed only the first selected file, and an idle firmware state was treated as proof that the physical bed was empty.

Impact

Printer progress, discovery counts, offline state, bed-clearance safety, and batch queueing now behave consistently with the real four-printer hotspot setup.

Validation

• npm run check
• 41 unit tests
• Full simulator integration flow: upload, Auto-Print, pause, resume, cancel
• Live four-printer discovery verified on the Windows hotspot