Certain padding values cause plots to overtake all other subplots

Related to #4124 there is a set of padding values that will produce strange behavior such that a single subplot takes over all others.

Some setup:

import numpy as np
import plotly.graph_objects as go
from plotly.subplots import make_subplots
line_x = np.arange(10)
line_y = line_x**2

What works:

n_subplots = 6

full_fig = make_subplots(
    rows=n_subplots,
    cols=1,
)

for row_idx in range(1, n_subplots+1):
    full_fig.add_trace(go.Scatter(
            x=line_x,
            y=line_y,
            showlegend=False,
        ), row=row_idx, col=1)

This produces 6 different subplots of the same line.

What doesn't work:

line_x = np.arange(10)
line_y = line_x**2

n_subplots = 8
padding_size = 0.1

row_heights = [0.1]*n_subplots

specs = []
for _ in range(n_subplots):
    specs.append([{'b': padding_size/2., 't': padding_size/2.}])

full_fig = make_subplots(
    rows=n_subplots,
    cols=1,
    row_heights=row_heights,
    specs=specs,
)

for row_idx in range(1, n_subplots+1):
    full_fig.add_trace(go.Scatter(
            x=line_x,
            y=line_y,
            showlegend=False,
        ), row=row_idx, col=1)

Adding the padding and row heights now generates only a single plot.

Start by running the provided make_subplots examples and compare the six-row case with the eight-row case using row_heights and specs padding. Trace how make_subplots handles row heights and top/bottom padding, then verify the fix against the reproduced configuration. Done means all eight subplots render separately rather than one subplot taking over.