Should LinearColormap.to_step(n=1) create a one-step colormap?

从 branca/colormap.py 中 LinearColormap.to_step 附近开始，这里的 n - 1.0 计算会在 n=1 时引发错误，并重现 issue 中的失败。根据现有 API 和周围的测试，确定单步操作的预期行为；当所选行为已实现、有明确说明，并由回归测试覆盖时，即视为完成。

LinearColormap(...).to_step(n=1) raises a ZeroDivisionError.

In map/choropleth workflows, after filtering or classification, there may be only one bucket/class to display, and a one-step colormap seems semantically reasonable.

Currently, we get:

❯ uv run --with branca python
Python 3.12.3 (main, Mar 23 2026, 19:04:32) [GCC 13.3.0] on linux
Type "help", "copyright", "credits" or "license" for more information.
>>> import branca
>>> from branca.colormap import LinearColormap
>>> print(branca.__version__)
0.8.2
>>> 
>>> cm = LinearColormap(["red", "blue"], vmin=0, vmax=1)
>>> cm.to_step(n=1)
Traceback (most recent call last):
  File "<stdin>", line 1, in <module>
  File ".../branca/colormap.py", line 450, in to_step
    index[i] * (1.0 - i / (n - 1.0)) + index[i + 1] * i / (n - 1.0),
                      ~~^~~~~~~~~~~
ZeroDivisionError: float division by zero

The API says n is the expected number of colors in the output StepColormap. When no index/data is supplied it constructs a regular grid from n; later, it computes colors using n - 1.0, which fails at n=1.

Possibilities:

• Could to_step(n=1) to return a StepColormap with one color/bin?
• Or should n <= 1 be explicitly rejected with a clear ValueError?
• Or a mention in the docs stating that callers need to special-case single-class data?