Add compression ratio estimation without full pipeline run

[vergoodigi]

Before running the full 14-stage pipeline, it would be useful to get a quick estimate of expected compression ratio. Something like:

estimate = engine.estimate(text)  # fast, <1ms
# {"estimated_reduction_pct": 45, "recommended_stages": ["ionizer", "rle"]}

This could use Cortex content detection + heuristics from historical benchmarks, without actually running compression.