Use microbatch B>1 not possible

[LeoBuron]

Tracking issue for the work needed to enable microbatch dimension B > 1 in ODT.

Today the framework operates with implicit B = 1: each microbatch is a single sample, and shapes like [1, F] are common. Several places encode this assumption explicitly or implicitly:

• Loss-API uniform-microbatch contract (docs/CONVENTIONS.md → "Loss API: microbatch contracts" — landed with Loss API: unify forward+backward reduction semantics for microbatching with batch dimension #135)
• DataLoader returns one sample per batch index in getBatch — shape [1, ...] not [B, ...]
• Layer forward/backward implementations may not support batched input
• Training-loop per-microbatch counters assume B = 1

When B > 1 becomes a real feature target, the sub-issues below capture the verified-needed work. New sub-issues land here as we discover them — this is the umbrella, not the implementation bag.

Sub-issues

• Microbatch shape contract asserts in loss forward/backward #153 — Microbatch shape contract asserts in loss forward/backward

Out of scope

This issue does not get implemented as a single PR. It is a tracking umbrella; each sub-issue lands independently.