Clarity on whether supported legacy name aliases are supported properties

[gsnedders]

1. should CSS.supports("word-wrap", "break-word") return true or false?
2. should CSS.supports("(word-wrap: break-word)") return true or false?
3. should "word-wrap" in window.getComputedStyle(document.documentElement) be true or false?
4. should [...window.getComputedStyle(document.documentElement)].includes("word-wrap") be true or false?

IMO:

1. all UAs say true, which falls out from "for string arguments"
2. all UAs say true, which falls out from converting it "at parse time" (but we may want to be more explicit that we mean any property name parsing, because "at parse time" is kinda unclear as to what parse time)
3. all UAs say true; but this I'm unsure about what the current specs imply: does "applies in the CSSOM [for] property accessors" effect the generation of WebIDL attributes?
4. all UAs supporting word-wrap as a legacy name alias say false; does this just fall out from the getter giving "longhand properties that are supported CSS properties", on the basis that they aren't supported properties (they're instead legacy name aliases)?

[tabatkins]

Definitely true for 1 and 2, because both of those successfully parse.

Unsure about 3 and 4.