`make_column_metadata` should prioritize spec-defined lengths over auto-detected lengths

[trinathpanda]

Problem

The function make_column_metadata automatically infers the maximum length of variables from the row-level dataframe. While this works dynamically, it can create conflicts when the specification (e.g., SDTM/ADaM spec or Define-XML) already prescribes a fixed length for certain variables. The inferred length may not match the required length in the specification, leading to inconsistencies.

Expected Behavior

• If a variable length is explicitly defined in the specification, the function should honor the spec-defined length.
• Auto-detection of maximum length should only apply when no specification length is provided.

Actual Behavior

• The function always auto-detects the maximum length, even if the specification provides a fixed length.
• This can result in mismatched metadata compared to the standard specification.

Example

• Specification: --TERM length = 200
• Row-level data max length detected = 87
• Generated metadata sets length = 87 (instead of 200 from the spec).

Impact

• This causes downstream misalignment with Define-XML and potential compliance issues in regulatory submissions.
• Suggested Fix
• Update make_column_metadata to:
• Check for a specification-defined length first.
• Fall back to auto-detection only if no length is defined in the spec.