Update figA_iwa_learn_vs_frozen_a0667 (full run9 curve)

Update appendix training-dynamics figures (GroundFlow only, SD-RPN removed) + nips.tex + plot script

Add appendix training-dynamics figures (figA_loss_{4b,8b}, figA_gn_{4b,8b}, figA_iwa_learn_vs_frozen_a0667)

source_grid_oblique: shorter sides (flatter, 720x280)

Flatter lying-down perspective for source_grid_oblique

textvqa patch tiles: switch to camera image; ROI + SINK + attention figures (PNG)

32B full results: nips.tex (Vanilla/SD-RPN/GroundFlow), trainer_state for run2 + SD-RPN, textvqa patch tile figures (PNG), eval/sync scripts

sync figures from /opt/tiger/thothvl_pretrain (HF tokens redacted)