refactor: align task groups to paper (JOBSIM/SKILLSIM → Semantic Similarity + Candidate Ranking)

[Mattdl]

What

Aligns the codebase task grouping with the paper's 7 task groups. Previously, the code had JOBSIM (Job Title Sim + Candidate tasks) and SKILLSIM (SkillMatch-1K), which didn't match the paper's "Semantic Similarity" and "Candidate Ranking" groups. The freelancer matching datasets are being updated from JOBSIM → CANDIDATE_RANKING.

Changes

Enum refactoring:

• JOBSIM & SKILLSIM → SEMANTIC_SIMILARITY (Job Title Similarity + SkillMatch-1K)
• JOBSIM → CANDIDATE_RANKING (for Query-Candidate + Project-Candidate matching)

Files changed:

• src/workrb/tasks/abstract/ranking_base.py — Updated RankingTaskGroup enum
• src/workrb/tasks/ranking/job_similarity.py — JOBSIM → SEMANTIC_SIMILARITY
• src/workrb/tasks/ranking/skill_similarity.py — SKILLSIM → SEMANTIC_SIMILARITY
• src/workrb/tasks/ranking/freelancer_project_matching.py — JOBSIM → CANDIDATE_RANKING
• tests/test_evaluate_multiple_models.py — Updated hardcoded group name strings
• README.md — Reorganized task table by task group, updated aggregation example

Task group mapping (paper ↔ code)

Paper	Code Enum
Job-to-Skill Matching	JOB2SKILL
Skill-to-Job Matching	SKILL2JOB
Job Normalization	JOB_NORMALIZATION
Skill Normalization	SKILL_NORMALIZATION
Semantic Similarity	SEMANTIC_SIMILARITY (new)
Skill Extraction	SKILL_EXTRACTION
Candidate Ranking	CANDIDATE_RANKING (new)