Unexpected margin-left behavior in datatables with fixed row

Dash Python
1

When using a fixed header on a datatable, the margin-left css attritube switches to -1.25px for the whole table (compared to the same table without fixed header), hiding the left side border.
On top of that, the margin-left attribute for the header row oscillate rapidly between 0 and 1.25 as I scroll down the table. This result in the left side border of the header row blinking rapidly.

I have made a short video to illustrate the issue.

First I show you what the table should look like with margin-left set to 0 (this is how it looked before enabling fixed header). As you can see, simply putting the mouse over the table reset the attribute to the incorrect -1.25px value.
I then highlight the property that “flickers” between 0 and -1.25 px and randomly scroll through the table to show the issue.

Here is the code for this table:

def today_eqs():
    master_df = database.get_unfiltered_df().sort_values("time", ascending=False)

    df = master_df[master_df["date"] == dt.utcnow().date()][["time", "mag", "depth", "coordinate", "energy"]]

    return dash_table.DataTable(
        id='today_eq',
        columns=[
            dict(id='time', name='Datetime (UTC time)', type='datetime'),
            table_styling.ColumnFormat("depth", "Depth").depth(),
            table_styling.ColumnFormat("mag", "Magnitude").magnitude(),
            table_styling.ColumnFormat("energy", "Energy release (in joules)").energy(),
            dict(id='coordinate', name='Coordinate', type='text'),
        ],
        data=df.to_dict('records'),
        sort_action="native",
        fixed_rows={'headers': True},
        style_cell=dict(textAlign='left'),
        style_header=dict(
            backgroundColor='rgb(30,30,30)',
            color='white'
            ),
        style_data=dict(
            backgroundColor='rgb(50,50,50)',
            color='white'
            ),
        style_table={
            'width': '{}%'.format(100)
        },
        style_data_conditional=[
            table_styling.every_other_rows(df),
            *table_styling.highlight_max_value(df),
        ],
    )