Add Common Issues table to databricks-mlflow-evaluation skill

[CheeYuTan]

Summary

• Adds a 10-entry Common Issues troubleshooting table to the databricks-mlflow-evaluation skill
• Covers genai.evaluate() vs evaluate(), predict_fn signatures, dataset format, scorer issues, MemAlign alignment, and GEPA optimization requirements

Test proof

Tested MLflow evaluation patterns locally and against live workspace e2-demo-field-eng:

Test	Result
mlflow.genai import	PASS
MLflow version >= 3.x	PASS — mlflow version: 3.7.0
Correct dataset format (nested inputs)	PASS — verified {"inputs": {"query": "..."}} structure
@scorer decorator available	PASS — mlflow.genai.scorer importable
Built-in scorers importable (Correctness, Safety, Guidelines)	PASS — all 3 import successfully

TEST: MLflow genai import check
  PASS

TEST: MLflow version check (>= 3.x)
  PASS
  → mlflow version: 3.7.0

TEST: Scorer decorator available
  PASS
  → True

TEST: Built-in scorers importable
  PASS
  → [True, True, True]

All 5/5 MLflow evaluation tests passed.

[calreynolds]

Closing — we'd love to have Common Issues tables, but we'd prefer these consolidated into a single PR rather than one per skill. Feel free to resubmit as a single combined PR if you're up for it!