Fix #130: Expose delay fitting options in CLI config classes

[Devguru-codes]

Hello @ltorres6 sir, I am making this pr because i saw the issue created by you. This PR exposes the delay fitting options that already exist in the library layer (fit_model() in osipy/dce/fitting.py) through the CLI configuration system, allowing users to enable and configure arterial delay fitting via YAML config files. The changes made are -

1. osipy/cli/config.py - Added fit_delay: bool = False and delay_bounds: list[float] = [0.0, 60.0] to DCEFittingConfig, with a validator ensuring delay_bounds is a valid [lower, upper] pair.

2. osipy/pipeline/dce_pipeline.py - Added fit_delay and delay_bounds to the DCEPipelineConfig dataclass, and passed fit_delay=self.config.fit_delay in the fit_model() call inside DCEPipeline.run().

3. osipy/cli/runner.py - Wired fit_delay and delay_bounds from the YAML config through both runner code paths:

   • NIfTI path (_run_dce): Passed when building DCEPipelineConfig
   • DICOM path (_run_dce_from_dicom): Passed directly to fit_model()

4. tests/unit/cli/test_config.py - Added 5 new tests:

   • Default values for fit_delay and delay_bounds
   • Custom delay_bounds parsing
   • Validation for wrong-length and inverted bounds
   • Full YAML round-trip test with delay fitting enabled

DSC Note

DSC delay maps (Ta/Tmax) are already computed automatically by the SVD deconvolution methods - they come from the residue function peak, not from a configurable fitting toggle. No DSC config changes are needed.

Usage

Users can now enable delay fitting in their YAML config:

modality: dce
pipeline:
model: extended_tofts
fitting:
fit_delay: true
delay_bounds: [0.0, 30.0]

I will wait for suggestions from your side and then update this pr accordingly. Thank you.
Fixes #130