Error "KeyError: 'id'" on st.dataframe where style.format is used - sensitive to dataframe size

issue-fail.py

import streamlit as st
import pandas as pd

df = pd.read_csv("https://leg-2-lou-docs-pics.s3.us-west-2.amazonaws.com/reports/invoices_fail_8457_rows.csv", dtype=str, parse_dates=[8,9])

col1, col2 = st.columns(2)

with col1:
    st.markdown("# Invoices ")

with col2:
    st.metric("Row count", len(df))

st.markdown("## Invoice Data CSV from S3")

#Convert text to float
df['ItemSubTotalAmount'] = df['ItemSubTotalAmount'].astype(float)
df['TaxAmount'] = df['TaxAmount'].astype(float)
df['TotalAmount'] = df['TotalAmount'].astype(float)

# This next line always works
#st.dataframe(df);

# Applying formating fails based on dataframe size
st.dataframe(df.style.format({'ItemSubTotalAmount': '{:.2f}', 'TaxAmount': '{:.2f}', 'TotalAmount': '{:.2f}'}))

issue.py

import streamlit as st
import pandas as pd

df = pd.read_csv("https://leg-2-lou-docs-pics.s3.us-west-2.amazonaws.com/reports/invoices_success_8456_rows.csv", dtype=str, parse_dates=[8,9])

col1, col2 = st.columns(2)

with col1:
    st.markdown("# Invoices ")

with col2:
    st.metric("Row count", len(df))

st.markdown("## Invoice Data CSV from S3")

#Convert text to float
df['ItemSubTotalAmount'] = df['ItemSubTotalAmount'].astype(float)
df['TaxAmount'] = df['TaxAmount'].astype(float)
df['TotalAmount'] = df['TotalAmount'].astype(float)

# This next line always works
#st.dataframe(df);

# Applying formating fails based on dataframe size
st.dataframe(df.style.format({'ItemSubTotalAmount': '{:.2f}', 'TaxAmount': '{:.2f}', 'TotalAmount': '{:.2f}'}))

Run
streamlit run issue.py
or
streamlit run https://gist.github.com/evoshawkins/1432dc9dd3c6a4358c8eb28688519734/raw/c3dbddb868456966cf16222dc9ff3fd5a467e5f8/issue.py

The dataframe will load and format the 3 float columns with 2 trailing decimal places.

On the other hand
streamlit run issue-fail.py
or
streamlit run https://gist.github.com/evoshawkins/1432dc9dd3c6a4358c8eb28688519734/raw/c3dbddb868456966cf16222dc9ff3fd5a467e5f8/issue-fail.py

The dataframe will not load but instead give an error 'KeyError: ['id']'

The different between the two is that the CSV loaded from S3 in the 2nd case has one more row than the one that succeeds.