[ENH] Add UnconditionalDistfitRegressor and DeterministicReductionRegressor

[arnavk23]

Reference Issues/PRs

Towards #7

What does this implement/fix? Explain your changes.

This PR adds two unconditional baselines for probabilistic regression to improve benchmarking:

UnconditionalDistfitRegressor

• Fits a univariate distribution to y using distfit.
• Ignores X entirely by design (feature-agnostic baseline).
• Supports stable distribution choices (norm, laplace) and histogram mode.

DeterministicReductionRegressor

• Wraps a deterministic regressor.
• Uses deterministic predictions as location (mu) and constant uncertainty from training target variance:
• Gaussian: sigma = sqrt(var(y_train))
• Laplace: scale = sqrt(var(y_train) / 2)

Additional updates:

• Aligns behavior with current distfit compatibility constraints.
• Fixes probabilistic output contract behavior for quantiles/intervals and estimator checks.
• Cleans up deprecated KDE path handling (not treated as actively supported in this PR).

Limitations

UnconditionalDistfitRegressor

• Ignores features completely.
• Cannot model feature-conditional uncertainty or heteroscedasticity.
• Produces global uncertainty not adapted per sample.

DeterministicReductionRegressor

• Uses constant scale across all samples.
• Does not model input-dependent uncertainty.
• Both current y target univariate y.

Does your contribution introduce a new dependency? If yes, which one?

distfit is used as a soft dependency.

What should a reviewer concentrate their feedback on?

• Correctness of Gaussian/Laplace parameterization (mu, sigma, scale)
• Quantile/interval behavior and distribution output contracts
• Mean consistency between wrapped deterministic predictions and returned distributions
• Baseline semantics and documented limitations

Did you add any tests for the change?

Yes. Tests cover:

• Invalid distribution type handling

• Univariate vs multioutput behavior

• Distribution parameter correctness (loc/scale, sigma/scale)

• Mean consistency with deterministic predictions

• Estimator-wide contract checks for probabilistic outputs

Any other comments?

KDE-related behavior was cleaned up to reflect upstream instability/deprecation in the distfit/SciPy stack. This PR does not present KDE as an actively supported option.

• Ensures compatibility with distfit 2.0.1: default distr_type is now 'norm' (was 'best').

PR checklist

For all contributions

• I've added myself to the list of contributors with any new badges I've earned :-)
  How to: add yourself to the all-contributors file in the skpro root directory (not the CONTRIBUTORS.md). Common badges: code - fixing a bug, or adding code logic. doc - writing or improving documentation or docstrings. bug - reporting or diagnosing a bug (get this plus code if you also fixed the bug in the PR).maintenance - CI, test framework, release.
  See here for full badge reference
• The PR title starts with either [ENH], [MNT], [DOC], or [BUG]. [BUG] - bugfix, [MNT] - CI, test framework, [ENH] - adding or improving code, [DOC] - writing or improving documentation or docstrings.

For new estimators

• I've added the estimator to the API reference - in docs/source/api_reference/taskname.rst, follow the pattern.
• I've added one or more illustrative usage examples to the docstring, in a pydocstyle compliant Examples section.
• If the estimator relies on a soft dependency, I've set the python_dependencies tag and ensured
  dependency isolation, see the estimator dependencies guide.

[arnavk23]

Updated the UnconditionalDistfitRegressor docstring doctest expectation to match current return contract and stable scalar formatting in unconditional_distfit.py.
Hardened plotting test backend selection to avoid tkinter backend issues by forcing Agg in test_proba_basic.py.