Bars object does not respect min or max of x-LinearScale correctly

[andymcarter]

Describe the bug
When using the Bars object together with a LinearScale, the min and max attributes of the scale does affect the chart plotting areas, but the precise min and max values are not respected.

The behaviour appears to be correct for a y-axis LinearScale.

To Reproduce
import bqplot as bqp
x_sc = bqp.LinearScale(min=0)
y_sc = bqp.LinearScale()
ax_x = bqp.Axis(scale = x_sc)
ax_y = bqp.Axis(scale = y_sc, orientation='vertical')

bars = bqp.Bars(x = [10,12,14], y = [10,12,14], scales={'x': x_sc, 'y': y_sc})

fig = bqp.Figure(marks = [bars], axes = [ax_x, ax_y,])
fig

Expected behavior
The chart x axis should start from 0.

Screenshots

Environment (please complete the following information):

• macOS, Chrome, bqplot 0.11.5

Additional context
Add any other context about the problem here.

[marketneutral]

I am also observing that LinearScale does not respect min or max in the pyplot API. Example:

import pandas as pd
import numpy as np
from datetime import datetime
from bqplot import pyplot as blt
from bqplot import LinearScale
from ipywidgets import HBox

def event_handler(name, value):
    sel = price_line.selected
    if sel is None or len(sel) < 2:
        pass
    else:
        zoomed_ts = ts.iloc[sel[0]:sel[-1]:, ]
        zoom_line.y = zoomed_ts.values
        zoom_line.x = zoomed_ts.index.values

start_date = datetime(2000,1,1)
end_date = datetime.today()
dates = pd.date_range(start_date, end_date, freq='M')
ts = pd.DataFrame(data=np.random.rand(len(dates),1), index=dates, columns=['XYZ']).cumsum()

price_fig = blt.figure()
price_line = blt.plot(ts.index.values, ts.values)
blt.axes()['x'].tick_format = '%b-%Y'
blt.axes()['x'].num_ticks = 5
brush = blt.int_selector(event_handler)

zoom_fig = blt.figure()
zoom_line = blt.plot(ts.index.values, ts.values)
zoom_line.scales['y'] = LinearScale(min=0, max=100)

The y-axis on the plot on the right should not exceed 100.

I also expected the axis to be fixed at 0 to 100, as it would be in matplotlib when you set the range of the axis.

[thirdreplicator]

As a work-around you can use:

plt.ylim(0, 100)