Feature: multiple files, channels or choices as one typed input #60

Goal

Support scientific tasks that naturally accept several input files, several selected image channels, or a list of experimental conditions, without requiring comma-delimited strings or custom ad hoc JSON.

Scope and cautions

• Investigate declarative typed sequences: e.g. list[File], list[Image] or an explicit multi-selection marker for short Literal/Enum options. Specify which is preferable for backward compatibility.
• Specify selection order, optional/required, maximum count, empty selection and per-image calibration/provenance.
• A collection of multiple images provided to a single function is different from host-level batch execution of a single-image tool; do not conflate them.
• Each host chooses a suitable multiple-file/layer/control UI. Hosts lacking a custom widget must still supply the same validated serialized input.
• Avoid implicit merging of time axes, channels or pixel grids.

Acceptance criteria

• A tool receives ordered selections faithfully through CLI/Notebook/Napari/Fiji/QuPath.
• Per-item validation and clear errors for missing/unreadable inputs.
• No free-text fallback that silently changes the requested list.
• Tests cover empty, one, several inputs, mismatched calibrations and concurrent file selection.

Priority: Medium.