Tighten and prune BTE lookup query templates

[maximusunc]

Several active multi-hop templates under-constrained their edges, causing the BTE lookup worker to return excessive results. Tighten the four over-broad Drug-treats-Disease templates:

• Chem-affects-Gene-has_phenotype-DoP: constrain the bare affects first hop with an activity_or_abundance aspect qualifier (direction-agnostic).
• Chem-affects_increases-Gene-loss_function-Disease and its _decreases mirror: the gene->disease edge had no predicate at all; give it associated_with + a contributes_to qualified predicate (following the existing associated_variant template pattern), and add an activity_or_abundance aspect to the first hop.
• Chem-treats-PhenoOfDisease: narrow the catch-all treats_or_applied_or_studied_to_treat predicate to treats/ameliorates.

Also remove 14 template files that no group in template_groups.json references (the old_ideas/ and NonChem-treats-Disease/ directories plus several loose correlation/interaction one-hops), so they no longer sit as unreachable dead weight. The 10 referenced templates are unaffected.

After making these changes, I ran the Test Harness against my local BTE and here are the results:

Local, BTE ran all 85 queries in 4:38:

{
  "bte": {
    "TopAnswer": {
      "PASSED": 59,
      "FAILED": 41,
      "NO_RESULTS": 2,
      "SKIPPED": 0,
      "ERROR": 0
    },
    "Acceptable": {
      "PASSED": 38,
      "FAILED": 50,
      "NO_RESULTS": 9,
      "SKIPPED": 2,
      "ERROR": 0
    },
    "BadButForgivable": {
      "PASSED": 12,
      "FAILED": 1,
      "NO_RESULTS": 4,
      "SKIPPED": 0,
      "ERROR": 0
    },
    "NeverShow": {
      "PASSED": 379,
      "FAILED": 21,
      "NO_RESULTS": 35,
      "SKIPPED": 0,
      "ERROR": 0
    }
  }
}

vs CI (and I ran the same locally to check parity and it's spot on, and to check timing and it took 17:24):

{
  "shepherd-bte": {
    "TopAnswer": {
      "PASSED": 56,
      "FAILED": 44,
      "NO_RESULTS": 2,
      "SKIPPED": 0,
      "ERROR": 0
    },
    "Acceptable": {
      "PASSED": 40,
      "FAILED": 48,
      "NO_RESULTS": 9,
      "SKIPPED": 2,
      "ERROR": 0
    },
    "BadButForgivable": {
      "PASSED": 10,
      "FAILED": 3,
      "NO_RESULTS": 4,
      "SKIPPED": 0,
      "ERROR": 0
    },
    "NeverShow": {
      "PASSED": 353,
      "FAILED": 47,
      "NO_RESULTS": 35,
      "SKIPPED": 0,
      "ERROR": 0
    }
  }
}

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 53.08%. Comparing base (e39a05f) to head (51ef722).
⚠️ Report is 7 commits behind head on main.
see 4 files with indirect coverage changes

---

Continue to review full report in Codecov by Harness.

Legend - Click here to learn more
Δ = absolute <relative> (impact), ø = not affected, ? = missing data
Powered by Codecov. Last update c0afa53...51ef722. Read the comment docs.

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.