testcase fixes in css/CSS2/borders

[faceless2]

Two changes:

1. border-N-width-applies-to-008.xht tests did not have a reftest and were slightly dependent on default font-metrics (specifically, nbsp being 0.25em wide). Redid to use an image, although they're still testing the same thing. Added a matching reftest

2. border-color-011.xht was incorrect, result of test has reversed since the rules regarding "currentColor" inheritance were clarified in css-color-4

3. additional whitespace in bidi-003.xht was causing reftest mismatch

[TalbotG]

@faceless2 (Mike),

In your border-*-width-applies-to-008 tests:
Shouldn't you be saying that it's a square, not a rectangle?

line 26 or 27: <img src="support/1x1-white.png" />
is going to trigger a validation error. Same validation error in your border-n-width-applies-to-008-ref.xht reference file. You need to wrap the img element with a block like a <div>:

Line 27, Column 43: document type does not allow element "img" here; missing one of "p", "h1", "h2", "h3", "h4", "h5", "h6", "div", (...)

Also,
alt="Image download support must be enabled"
is missing. We want to know easily and quickly if there are any broken links or incorrect linkage to images.

[TalbotG]

In your border-*-width-applies-to-008 tests:

Since img is used, then

line 18: display: inline;

can be safely removed too.

[svgeesus]

The change to border-color-011.xht correctly captures the currentColor changes in css-color-4 (and css-color-3, edited rec).

[faceless2]

Closed; reapplied as single patches.

• see 'currentColor' is now inherited as a keyword #21662
• see Bidi whitespace collapse was preventing reftest match #21663
  ... and have left border-N-width-applies-to-008.xht - this is an old test and the change was was significant, probably not worth it.