fix: raise ValueError for h>1 in forecast_fitted_values

[GauravPatil2515]

Fixes #1032

Description

This PR addresses the issue where forecast_fitted_values silently ignores multi-step expectations or behaves ambiguously when users try to generate in-sample forecasts for h > 1.

Forecast fitted values are conceptually 1-step-ahead predictions. To prevent user confusion and incorrect usage, this change makes the restriction explicit.

Changes

• Core Logic: Updated _StatsForecast.forecast_fitted_values and StatsForecast.forecast_fitted_values to accept an h parameter (defaulting to 1).
• Validation: Added a check to explicitly raise a ValueError if h > 1. The error message guides new users to use cross_validation or forecast() for multi-step predictions.
• Tests: Added regression tests (tests/test_issue_1032.py) to verify:
  • h=1 works as expected (default logic preserved).
  • h>1 correctly raises ValueError.

Checklist

• Code passes all tests
• New tests added
• Docstrings updated to reflect the new h parameter and ValueError

[CLAassistant]

Thank you for your submission! We really appreciate it. Like many open source projects, we ask that you sign our Contributor License Agreement before we can accept your contribution.
_{You have signed the CLA already but the status is still pending? Let us recheck it.}