add a "step" or "precision" element to the number formats

[Tieske]

The current format for numbers accepts min and max, but no precision.

Can we make the format look like: min:max:step ?

• default would be :; neither min nor max specified
• only a min: 10:, 10 or greater
• only a max: :20 20 or smaller
• only a step size: ::2 then the base is 0, so in the example, is 0, 2, 4, etc are allowed, 1, 3, 5, etc are invalid.
• specify them all: 10:20:2, then remainder((max - min)/step) == 0 must hold to validate

Benefits:

• Allows predictable rounding by the device if a value is not matching the step size. Since it is currently unclear, rounding should probably be required to stay in line.
• Provides a hint to the UI as to how to render the value. For example a thermostat with a setpoint in 0.5 degrees precision. Now if the value is set to 18, with a format 5:30:0.5, the UI knows it should display with 1 decimal precision. So despite 18 being an integer, it can display 18.0.

Examples:

• seconds with millisecond precision: 0::0.001
• temperature with 0.5 degrees precision: 5:30:0.5
• GPS locations/distance with 10 meter precision: ::10

[Tieske]

based on #216 ; the proposal above specifies the min+max as "inclusive"

[Tieske]

Not supported case is "only odd numbers" (no min/max). This forces "base" to be non-0.

[Tieske]

In the case that neither min nor max is specified, the current value of the property can be used to determine the base.