Increase color stop index for transparent edges.

[brendandahl]

The current number is too small for chrome and causes issues.

Fixes #9664

[brendandahl]

/botio test

[pdfjsbot]

From: Bot.io (Windows)

---

Received

Command cmd_test from @brendandahl received. Current queue size: 0

Live output at: http://54.215.176.217:8877/a3f8ad437bbb3f2/output.txt

[pdfjsbot]

From: Bot.io (Linux m4)

---

Received

Command cmd_test from @brendandahl received. Current queue size: 0

Live output at: http://54.67.70.0:8877/0e1d8bcd72e8cc8/output.txt

[pdfjsbot]

From: Bot.io (Windows)

---

Failed

Full output at http://54.215.176.217:8877/a3f8ad437bbb3f2/output.txt

Total script time: 29.07 mins

• Font tests: Passed
• Unit tests: Passed
• Regression tests: FAILED

Image differences available at: http://54.215.176.217:8877/a3f8ad437bbb3f2/reftest-analyzer.html#web=eq.log

[pdfjsbot]

From: Bot.io (Linux m4)

---

Failed

Full output at http://54.67.70.0:8877/0e1d8bcd72e8cc8/output.txt

Total script time: 35.90 mins

• Font tests: Passed
• Unit tests: Passed
• Regression tests: FAILED

Image differences available at: http://54.67.70.0:8877/0e1d8bcd72e8cc8/reftest-analyzer.html#web=eq.log

[brendandahl]

If anyone wants to play with this further...try out https://codepen.io/anon/pen/xJQXRJ and adjust the 2nd and 2nd to last color stops.

[brendandahl]

I was talking with Patrick and he mentioned the values are probably getting packed into a byte hence 254/255 = 0.996 so above .996 gets rounded to 1.

[brendandahl]

We should also test if this is actually an issue on the bots since they may not be using skia gl.

[timvandermeij]

@brendandahl I notice some small regressions in the reference test results. Do you want to continue this PR or should it be revisited?

[timvandermeij]

Closing since the underlying ticket is resolved by an upstream fix.