Exclude unknown work time from part-time searches

[emnl51]

Summary

• inspect the title, available description, weekly-hours/workload text, and provider metadata before classifying work time
• hard-exclude unresolved Work time unknown results from part-time and working-student searches even in preference mode
• hide previously stored unknown-work-time rows from those profiles immediately after deployment
• keep confirmed Werkstudent, Teilzeit, Minijob, Nebenjob, Studentenjob, hour, and workload signals
• apply the rule consistently to dedicated Search Jobs, the legacy/global search path, review lists, details, and notifications
• update README and the unreleased changelog

Verification

• 246 pytest tests passed
• Ruff check and format check passed
• all JavaScript files passed node --check
• clean diff check

[chatgpt-codex-connector]

💡 Codex Review

Here are some automated review suggestions for this pull request.

Reviewed commit: 35ffa384d7

ℹ️ About Codex in GitHub

Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you

• Open a pull request for review
• Mark a draft as ready
• Comment "@codex review".

If Codex has suggestions, it will comment; otherwise it will react with 👍.

Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".

[chatgpt-codex-connector]

Use the scoring classifier when filtering saved jobs

Confirmed vacancies are hidden when their work time is expressed only as weekly hours, workload percentage, PART_TIME provider metadata, or signals such as Student Assistant: assess_employment_fit recognizes these as part-time and can score/notify them, but classify_job_metadata still returns employment_type == "unknown". This post-query check consequently removes valid matches from the review queue; the filtering decision should reuse the employment assessment or make the metadata classifier recognize the same confirmation signals.

Useful? React with 👍 / 👎.

[chatgpt-codex-connector]

Filter unknown-work rows before applying LIMIT

When an upgraded database contains legacy non-excluded rows with genuinely unknown work time, the query selects the newest limit rows before this loop discards them. For example, 100 recent unknown rows can make the default API response empty even when older confirmed part-time matches exist, with no offset-based way for the UI to reach them. Apply the exclusion in the query or keep fetching until the requested number of eligible rows is collected.

Useful? React with 👍 / 👎.