Add time-series-forecasting task schema

[pjhul]

Scaffolds the new time-series-forecasting task with input/output JSON schemas, generated TypeScript types, task metadata, and registry wiring. pnpm check, test, and lint:check on the tasks package all pass locally.

Design decisions worth flagging

1. inputs is a list of structured series objects, not a string or binary payload. This diverges from existing HF tasks but aligns closely with mature time-series libraries (Darts, GluonTS, AWS SageMaker Chronos). Parallel-array encodings don't survive contact with ragged lengths or per-series optional fields (timestamps, covariates, metadata).

2. target is always 2D [num_timesteps][num_channels]. Univariate series use a length-1 channel axis. Avoids oneOf 1D/2D ambiguity at length-1 inputs.

3. Missing observations are null inline inside target, not a separate observed_mask. Matches pandas, R, SQL, Arrow convention. HF's TimeSeriesTransformer uses past_observed_mask because PyTorch tensors can't hold null — but for API consumers dealing with a separate mask is much more cumbersome than in-line null's.

4. Input uses start + parameters.frequency rather than an explicit timestamps array. Matches the Darts TimeSeries model.

5. quantile_predictions is an array of {level, values} objects rather than a dict keyed by string-float. Matches HF's pattern for enumerated scored outputs (text-generation logprobs, classification scores, fill-mask tokens). AWS Chronos uses a dict-keyed-by-string; we diverge intentionally for HF-ecosystem consistency.

6. Three uncertainty channels on output — mean (required), quantile_predictions, samples. All optional; they may coexist. No parametric distribution parameters. samples is the universal fallback for any distributional output.

Validation

• pnpm --filter tasks-gen inference-codegen regenerates clean inference.ts from the JSON schemas
• pnpm --filter @huggingface/tasks check (tsc)
• pnpm --filter @huggingface/tasks test
• pnpm --filter @huggingface/tasks lint:check
• Registered in packages/tasks/src/tasks/index.ts

---

Note

Low Risk
Primarily additive schemas/types and task metadata with minimal impact on existing tasks; risk is limited to potential downstream type/API expectations for the new task registration.

Overview
Adds a new time-series-forecasting task, including input/output JSON schemas, generated TypeScript inference types, task documentation (about.md), and curated task metadata (data.ts).

Wires the task into packages/tasks/src/tasks/index.ts by exporting the new inference types and enabling the task page/registry entry (switching it from undefined to getData(...)).

^{Reviewed by Cursor Bugbot for commit 5da52ea. Bugbot is set up for automated code reviews on this repo. Configure here.}