srcset parser should look for 'h' descriptors

[zcorpan]

Since we anticipate adding a h descriptor in the future (#85 #86) I think we should make the parser do the right thing now so that we don't get into a situation where Web compat expects 2x 100h to not drop the entire candidate (like we currently do with 2x 100w).

Step 12, add/change
[[
...
Let future-compat-h be absent.
...
(for "x" token)
If width, future-compat-h and density are not all absent, then let error be yes.
...
If the token consists of a valid non-negative integer followed by a "h" (U+0068 LATIN SMALL LETTER H) character
If future-compat-h and density are not both absent, let error be yes.
Apply the rules for parsing non-negative integers to the token. If the result is zero, let error be yes. Otherwise, let future-compat-h be the result.
]]

The only change here would be that specifying both h and x makes the candidate be dropped. The h descriptor is otherwise ignored like today.

[yoavweiss]

Doesn't that mean that we need to be able to predict whatever descriptors we may want to add in the future?

[zcorpan]

No, we just predict that h is likely to be added in the future and that it's likely we want it to have the same parsing behavior as w. For anything else we don't know anything and the default behavior of dropping unknown descriptors seems reasonable to me.