[FEATURE]: Heatmap with custom colour for NaN values

Description

Currently, z-values of NaN get rendered as transparent in heatmaps. Adding a new parameter nancolor to the heatmap would fix this by allowing users to set a specific colour for NaN values.

Why should this feature be added?

Biomedical data often contains missing values, visualising them explicitly within heatmaps (i.e. without imputation or defaulting to transparent which often visually clashes with the used colour scales) allows for proper communication of the data.

Mocks/Designs

See #1170

Notes

This feature has already been requested in in #975 and a PR with the necessary changes had previously been opened in #1170. While a workaround has been suggested, I believe this could hurt performance for large heatmaps and a new parameter would be more clean. Since the issue is still of concern, I created this new issue as requested here.

首先跟踪 plotly.js 中的热力图渲染入口，并查看 #1170 中讨论过的先前实现。确认当前 NaN z-values 的渲染方式，然后定义覆盖范围，证明在不改变正常热力图颜色或行为的情况下，会应用所提供的 nancolor。