Phase 4 — sharded tensor contractions across chips

[scttfrdmn]

Roadmap phase tracker

This issue tracks trntensor's work on Phase 4 of the trnsci roadmap.

See the suite-level roadmap
for the full phase matrix and cross-project dependencies. A reader-friendly
version lives at trnsci.dev/roadmap/.

What this phase means for trntensor:
THC / Tucker tensors compressing high-rank chemistry integrals
can themselves exceed single-chip HBM. Phase 4 shards the core
tensor and factor matrices across NeuronCores.

Done means:

• Sharded einsum: declare a tensor as sharded along one mode, contraction handles the reduction automatically.
• Sharded CP / Tucker decomposition — ALS iterations run cross-chip.
• Large-basis DF-MP2 via einsum path (complement to trnblas Phase 4).
• Benchmarks at trnsci.dev/trntensor/benchmarks/.

Coordination

• Label: phase-4-multichip — matches the same label in every sub-project.
• Cross-project dependencies, if any, are called out in the "Done means"
  list above. Link any PRs / child issues here for tracking.

Close this issue when all "Done means" items are satisfied.