[add] bias-var trade-off in resampling: make clearer / more explicit

[lisa-wm]

➕ ➕ ➕

⚠️ Please provide specific & actionable issue titles!

Chapter name / context

Performance evaluation, resampling

Description

• Conflicting goals: large train sets (for low bias) and large test sets (for low variance) ➡️ make crystal-clear why we need resampling and how we balance conflicting goals; and what we average over when aggregating over folds / repeating CV
• Explain: what's the problem with estimating variance in CV?
• Explain: what's the problem with LOO-CV?

Additional sources

...