refactor(recipe)!: backport vrnvl72 rename to r0.7.0 (#6419)

[malay-nagda]

Backport of #6419 to r0.7.0, cherry-picked from b1bb0a97ec5aad323eb2dd666e3f69aaebc1daff without conflicts. The backport patch ID matches the merged source commit.

Renames the Vera Rubin NVL72 GPU type and performance recipes from vr200 to vrnvl72. Existing --gpu vr200 and _vr200_ CLI recipe names remain deprecated aliases. Direct Python imports of the old modules and recipe functions must migrate to the new names.

Validation:

• All pre-commit hooks passed using uv run --no-project --with pre-commit pre-commit run --all-files. Standard project execution is blocked by the uninitialized Megatron-Core submodule in the isolated worktree.
• Focused torch-free launcher tests: 138 passed, 1 failed. test_known_cross_package_collisions_have_explicit_benchmark_precedence also fails on the unchanged r0.7.0 baseline with the same unregistered nemotron_3_super_pretrain_64gpu_gb200_bf16_config collision.
• git diff --cached --check passed.
• GPU-dependent tests were not run locally.

[github-actions]

Automatic Claude reviews have been retired. To request a pull-request review, post a comment containing:

/review

Add model=claude to use a Claude reviewer (the default is model=codex). mode=light|strict selects the review depth; for example, /review model=claude mode=strict. Comment /review help for all options.