Adds an option to fill the forecast segment with icons

[scinos]

(Comes from #669)

What problem does this PR solve?

Adds the option to add more than one icon per forecast span. This should improve readability, specially for wallboards.

Proposed solution

Adds a new config icon_fill with three possible values:

• single This is the default option and keeps current functionallity: it shows one single icon per span, centered.
• full Show one icon per forecast segment. For example, it will display 12 sun icons if the next 12h is sunny.
• <n> (an integer). Like full, but instead of displaying an icon per forecast segment, it will displaye an icon every nth segment. (In other words, full and 1 do the same). At the very least it will display one icon per span.

This is how it looks like:

Alternatively, we can have this as a boolean (icon_fill:false current behaviour, icon_fill:true fill all segments with icons), and introduce a second option icon_spacing (similar to label_spacing) to be able to space out the icons

[scinos]

@decompil3d I think I need some help with checking the config option icon_fill. I see there are config tests in two places:

• Class HourlyWeatherCard, method setConfig (https://github.com/decompil3d/lovelace-hourly-weather/blob/main/src/hourly-weather.ts#L192)
• Class HourlyWeatherCard, method renderCore (https://github.com/decompil3d/lovelace-hourly-weather/blob/main/src/hourly-weather.ts#L324)

Why do we need two? Which one should I use to test the validity of icon_fill option?

[decompil3d]

@decompil3d I think I need some help with checking the config option icon_fill. I see there are config tests in two places:

• Class HourlyWeatherCard, method setConfig (https://github.com/decompil3d/lovelace-hourly-weather/blob/main/src/hourly-weather.ts#L192)

• Class HourlyWeatherCard, method renderCore (https://github.com/decompil3d/lovelace-hourly-weather/blob/main/src/hourly-weather.ts#L324)

Why do we need two? Which one should I use to test the validity of icon_fill option?

I think checking validity in renderCore would be most consistent with the existing code. Thanks for checking.

[scinos]

@decompil3d added tests, this PR is ready for review now