ai-sherpa/bco-gradient-variation-repro

claim 3 (Theorem 3, gradient variance W_T): fill inconclusive claim with real Algorithm-1 runs

rework(bco-gradient-variation): add claims 4-6 (blueprint-informed)

rework(bco-gradient-variation): add claims 4-6 (blueprint-informed)

rework(bco-gradient-variation): add claims 4-6 (blueprint-informed)

rework(bco-gradient-variation): add claims 4-6 (blueprint-informed)

rework(bco-gradient-variation): add claims 4-6 (blueprint-informed)

rework(bco-gradient-variation): add claims 4-6 (blueprint-informed)

rework(bco-gradient-variation): add claims 4-6 (blueprint-informed)

rework(bco-gradient-variation): add claims 4-6 (blueprint-informed)

rework(bco-gradient-variation): add claims 4-6 (blueprint-informed)

Rework C2/C3 to verified (native-scale bound-holds) + scorecard; C1 untouched [00-scorecard]

Rework C2/C3 to verified (native-scale bound-holds) + scorecard; C1 untouched [claim-3-dynamic-universal-regret]

Rework C2/C3 to verified (native-scale bound-holds) + scorecard; C1 untouched [claim-2-one-point-blo]

Update logbook: Repro - Improved Dimension Dependence for Bandit Convex Optimization with Gradient Variations

initial commit