missing-data-bias-ehr.md's overall payer_code missingness is off (39.9% vs the real 39.6%)

[yakew7]

Where: explainers/missing-data-bias-ehr.md, the missingness table's payer_code row: 39.9% 37.5% 48.2% 10.7 pts.

Repro:

import pandas as pd
df = pd.read_csv('Healthcare Readmission/diabetic_data.csv')
print((df['payer_code'] == '?').mean() * 100)                              # 39.5574... -> 39.6%
print((df[df.race == 'Caucasian']['payer_code'] == '?').mean() * 100)      # 37.49 -> 37.5% (matches)
print((df[df.race == 'AfricanAmerican']['payer_code'] == '?').mean() * 100)  # 48.17 -> 48.2% (matches)

The two per-group figures (37.5%, 48.2%) and the 10.7-point gap are all correct - only the stated overall rate is wrong: 39.6%, not 39.9%. The weight and medical_specialty rows in the same table were independently checked and are exact.

Fix: "39.9%" -> "39.6%" in the payer_code row.