Figure.histogram: Improve parameters 'distribution' and 'distribution_pen' to draw distribution curves

[seisman]

In histogram, the -N option draws the distribution curve. It's in the form of -N[<mode>][+p<pen>], in which <mode> can be 0, 1, or 2 for different modes, and the option can be repeated when needed, and each curve can have different pen settings via +p.

Currently, -N option is aliased to distribution, so the syntax is like distribution="0+p1p,red", distribution=["0+p1p,red", "1+p1p,blue"].

This PR proposes a new syntax with two parameters distribution and distribution_pen:

GMT CLI	Old	New
-N	distribution=True	distribution=True*
-N1	distribution=1	distribution="median"
-N1+p1p,red	distribution="1+p1p,red"	distribution="median", distribution_pen="1p,red"
-N0+p1p,red -N1+p1p,red -N2+p1p,red	distribution=["0+p1p,red", "1+p1p,red", "2+p1p,red"]	distribution=["mean", "median", "lms"], distribution_pen="1p,red"
-N0+p1p,red -N1+p1p,blue -N2+p1p,yellow	distribution=["0+p1p,red", "1+p1p,blue", "2+p1p,yellow"]	distribution=["mean", "median", "lms"], distribution_pen=["1p,red", "1p,blue", "1p,yellow"]

*Supported for backward compatibility, but it's not recommended, because it's unclear which curve is drawn

Example

import numpy as np
import pygmt
from pygmt.params import Axis, Frame

rng = np.random.default_rng(seed=100)
mean = 100  # mean of distribution
stddev = 25  # standard deviation of distribution
data = rng.normal(loc=mean, scale=stddev, size=521)


fig = pygmt.Figure()
fig.histogram(
    data=data,
    frame=True,
    series=5,
    fill="red3",
    pen="1p",
    distribution=["mean", "median", "lms"],
    distribution_pen=["1p,blue", "1p,red", "1p,green"],
)
fig.show()