[nvfp4 training][rl] Add the four-over-six grouped GEMM and dispatcher wiring

[wolfcomos]

Stacked on #12/#13. Integrates the four-over-six recipe with MoE training for routed-expert layers, reusing existing pieces rather than adding kernels: the #13 quantizer (its CuTe DSL fast path dispatches automatically) and the existing F.scaled_grouped_mm for the GEMMs.

• Per-tensor activations: each token group's amax expands to a per-row vector so the whole packed tensor quantizes in one call — bitwise identical to quantizing each group separately — and one scaled_grouped_mm with per-group second-level scales runs the forward.
• Expert weights quantize in one flattened call for both block shapes (the quantizer accepts a per-row amax with 16x16 blocks under a tile-uniform contract): one kernel launch regardless of expert count.
• Row-scaled activations: one scaled_grouped_mm carrying the constant per-tensor factor in every group's slot, its bf16 output upcast and scaled by the raw per-row amaxes. The grouped GEMM emits bf16 only, which costs one output rounding vs per-group dense GEMMs; tests emulate that rounding in a loop oracle and compare bitwise (with a reduction-order SQNR fallback), and separately bound the rounding cost.
• Backward: high_precision or dequantized grouped GEMMs (four-over-six has no quantized backward); dequantized differentiates the quantized forward itself from the saved 4-bit codes and scales.

Wiring mirrors the Float8/MXFP8 recipes exactly: NVFP4FourOverSixTrainingOpConfig(TrainingOpBaseConfig) — a pytree-constant dataclass with explicit __eq__/__hash__ — plus an isinstance branch in _quantize_then_scaled_grouped_mm. The dispatcher branch owns the offs[-1] tail slice / zero-extend for padded token dispatchers that over-allocate A past the logical rows. quantize_() model conversion for the grouped config is future work; framework integrations drive the grouped GEMM dispatcher (see the torchtitan converters). Adds an NVFP4 four-over-six section to the moe_training README.

Tests: 40 grouped tests (grouped-vs-dense forward parity, row-scaled fused-vs-oracle, backward bitwise references including empty groups and ragged padding, dispatcher round-trip with over-allocated tails, fullgraph compile in both scale granularities) plus the dense suite with the relaxed quantizer combinations, all green on GB200.

Stack: #12 ← #13 ← this PR.

[wolfcomos]

Opened upstream as pytorch#4853 (same head). This fork draft stays as the staging copy.

[wolfcomos]

Live upstream as pytorch#4853 — closing the staging draft. The branch stays: it is the upstream PR's head.