fix(database): migrate valid_start_mjd / valid_end_mjd columns in caltables (fixes #102)

[jakobtfaber]

Summary

The caltables schema in dsa110_continuum/database/unified.py has included valid_start_mjd / valid_end_mjd in the CREATE TABLE for new databases, but the migration function _migrate_caltables_schema only added the provenance text columns (source_ms_path, solver_command, etc.). Existing pipeline.sqlite3 databases created before those columns therefore raised no such column: valid_start_mjd when the pipeline tried to register generated calibration tables (#102).

Changes

• dsa110_continuum/database/unified.py: add valid_start_mjd and valid_end_mjd to the new_columns list in _migrate_caltables_schema, guarded by the existing PRAGMA table_info probe. The CREATE TABLE and idx_caltables_valid index code are unchanged; the index is still created in ensure_calibration_database after the migration.

Verification

• make test-cloud PYTHON=/home/ubuntu/.venv-dsa110-312/bin/python passes (200 passed).
• ruff check dsa110_continuum/database/unified.py reports five pre-existing style/name issues in the file that are unrelated to this change (I001, F821 at query_df, and W291/W293 on migration SQL strings); I did not bulk-fix them per repo guidance.
• H17 validation pending: confirm that a real pipeline.sqlite3 with the old caltables schema successfully runs _migrate_caltables_schema and caltable registration no longer logs no such column: valid_start_mjd.

Closes #102.