feat: validate clinical attribute values against the 255-character limit (#87)

[officialasishkumar]

Summary

Clinical attribute values longer than 255 characters are silently truncated on load, with no feedback from either the validator or the loader. For example, in the aml_ohsu_2022 study the SURFACE_ANTIGENS_IMMUNOHISTOCHEMICAL_STAINS value for sample aml_ohsu_2022_2010_BA2304 is cut off mid-sentence, losing clinical detail.

The root cause is that clinical attribute values are stored in the clinical_patient.ATTR_VALUE and clinical_sample.ATTR_VALUE columns, both defined as varchar(255):

CREATE TABLE `clinical_patient` ( ... `ATTR_VALUE` varchar(255) NOT NULL, ... );
CREATE TABLE `clinical_sample`  ( ... `ATTR_VALUE` varchar(255) NOT NULL, ... );

Changes

• Add MAX_CLINICAL_ATTRIBUTE_VALUE_LENGTH = 255, mirroring the existing MAX_SAMPLE_STABLE_ID_LENGTH constant.
• In ClinicalValidator.checkLine() (the base class shared by SampleClinicalValidator and PatientClinicalValidator), raise an error when a value exceeds the maximum length, so the issue is caught during validation — before the data is loaded and truncated. This mirrors the existing 255-character check already applied to HGVSp_Short in the mutation validator.
• Add a unit test and a data fixture (data_clin_value_too_long.txt) verifying the boundary: a value of exactly 255 characters is accepted, while a 256-character value is rejected (reported with the correct line and column number).

This addresses the validator side of the issue, which is the recommended gatekeeper in the import workflow (validate → load): with validation enforcing the limit, the silent truncation is no longer reached through the normal pipeline.

Test plan

Ran the full validator test suite (as in CI, python:3.9):

./test_scripts.sh
Ran 183 tests
OK (skipped=1)

The new test:

ClinicalValuesTestCase.test_clinical_attribute_value_too_long

Fixes #87