docs: fix export-lora --lora-scaled syntax

[kkkzbh]

Overview

While setting up a LoRA workflow, I noticed that tools/export-lora/README.md no longer matches the current llama-export-lora CLI behavior.

The README documents --lora-scaled FNAME S, but the current help output and parser expect --lora-scaled FNAME:SCALE,....

This PR updates the README option description and multi-LoRA example to use the current syntax. No behavior is changed.

Additional information

This appears to have become stale after #18056 migrated repeated arguments to comma-separated syntax.

Testing

• Ran llama-export-lora --help
• Verified --lora-scaled foo.gguf 0.5 fails argument parsing with lora-scaled format: FNAME:SCALE
• Verified --lora-scaled foo.gguf:0.5 passes argument parsing and proceeds to model loading

Requirements

• I have read and agree with the contributing guidelines
• AI usage disclosure: YES
• I used Codex as an assistive tool to inspect the current CLI help/source and discuss this documentation-only update. I manually reviewed the README diff and understand the submitted change.