Color.isCSSColorString is buggy

[pixelzoom]

Color isCSSColorString was added in 4e26f60 for #1115.

This is affecting work in bamboo, and I'll need to work around it.

It's buggy, and these examples are failing:

Color.isCSSColorString( null )
Color.isCSSColorString( undefined )
Color.isCSSColorString( {} )

There's no reason that isCSSColorString should ever fail; it should return true or false.

For example:

> phet.scenery.Color.isCSSColorString( null )
Color.js:812 Uncaught TypeError: Cannot read property 'replace' of null
    at Function.preprocessCSS (Color.js:812)
    at Function.isCSSColorString (Color.js:831)
    at eval (eval at window.assertions.assertFunction (assert.js:25), <anonymous>:1:20)
    at window.assertions.assertFunction (assert.js:25)
    at Color.set (Color.js:90)
    at new Color (Color.js:47)
    at Function.toColor (Color.js:623)
    at <anonymous>:1:20

[pixelzoom]

Should Color.isCSSColorString( null ) -> 'transparent' ? That's essentially what PaintDef does.

[pixelzoom]

On Slack, @jonathanolson says that isCSSColorString is documented to only support string. If that's the case, please add an assertion. EDIT: That also seems like an unnessary restriction to me.

[jonathanolson]

Added an assertion. Presumably we could have a place checking to see if something is a valid color value if needed (and a type check included there), I'd prefer not to add type-checks in this type of function.