index.html's Tenant Screening card shows stale 7.17%/5.07%/29% numbers - the real script output and README.md both say 6.68%/5.16%/23%

[yakew7]

Where: index.html's entire "PROJECT - 07 - TENANT SCREENING" section (JSON-LD, homepage tickers, stats, terminal output, key insight, bias bars).

The gap: Running python3 "Tenant Screening/unfair.py" and python3 "Tenant Screening/fair.py" today prints:

Black Applicant High-Risk Flag Rate: 66.74%   White: 60.06%   Gap: 6.68%
95% CI: [3.99%, 9.33%]
...
Black Applicant High-Risk Flag Rate: 62.06%   White: 56.90%   Gap: 5.16%
95% CI: [2.43%, 7.88%]

which is exactly what README.md and Tenant Screening/README.md document: 6.68% → 5.16%, a 23% reduction. But index.html's entire Tenant Screening card still shows an older, internally-self-consistent-but-wrong set of numbers: 7.17% biased gap, 5.07% mitigated gap, 29% reduction, 67.05%/59.88% flag rates, 70.49%/63.50% model accuracy, 62.16%/57.08% mitigated flag rates, and CIs [4.50%, 9.81%]/[2.30%, 7.83%] - none of which match a fresh run of the scripts. These numbers have been there since the card was first added and were never updated when the rest of the repo was corrected to the current figures.

Repro:

python3 "Tenant Screening/unfair.py" | tail -6
python3 "Tenant Screening/fair.py" | tail -6
grep -n "7.17\|5.07\|29%\|67.05\|59.88" index.html

The script output (6.68%/5.16%, CI [3.99%,9.33%]/[2.43%,7.88%]) never matches any of the numbers grep finds in index.html.

Fix direction: Replace every occurrence of 7.17%/5.07%/29%/67.05%/59.88%/70.49%/63.50%/62.16%/57.08%/"[4.50%, 9.81%]"/"[2.30%, 7.83%]" in index.html's Tenant Screening card and both ticker mentions with the current 6.68%/5.16%/23%/66.74%/60.06%/62.06%/56.90% figures and their CIs, matching README.md and a fresh script run.

[yakew7]

Fixed in d071bd0: corrected every stale Tenant Screening number in index.html (JSON-LD, both homepage ticker mentions, project stats, both terminal outputs, key-insight text, all 4 bias bars, roadmap-meta line) from 7.17%/5.07%/29% to the real 6.68%/5.16%/23%, computed via a fresh run of Tenant Screening/unfair.py and fair.py.