More fixes in sycc422_to_rgb and sycc420_to_rgb when image width is odd

More fixes in sycc422_to_rgb and sycc420_to_rgb when image width is odd

This patch is supplementary to issue 418881

R=tsepez@chromium.org

Committed: https://pdfium.googlesource.com/pdfium/+/bff247d92c315f9aab2edb4aeaf0ef0b74ba321f

[Bo Xu, 2014-10-21 19:00:01]

bo_xu@foxitsoftware.com changed reviewers:
+ jun_fang@foxitsoftware.com, tsepez@chromium.org

[Bo Xu, 2014-10-21 19:00:01]

Hi Tom, this patch continues with 418881.

Openjpeg team mentioned the previous fix in 418881 is buggy.
When the last pixel is odd, it should be:

sycc_to_rgb(offset, upb, *y, *cb, *cr, r, g, b);
            ++y; ++r; ++g; ++b; ++cb; ++cr;

Previous patch was:

sycc_to_rgb(offset, upb, *y, *cb, *cr, r, g, b);
            ++y; ++r; ++g; ++b;

I am switching the style to match what they did regarding how to deal with the
even and odd number. In sycc420_to_rgb, it seems pretty lengthy, but it does
make it clearer to understand than using modulo trick. (I tried it but found
using modulo trick make it a little complicated due to the 2 for loop and the
operation on the last pixel).

Let me know what you think. Thanks.

[Tom Sepez, 2014-10-21 19:02:42]

You're right. LGTM. Thanks

[Bo Xu, 2014-10-21 19:15:45]

On 2014/10/21 19:02:42, Tom Sepez wrote:
> You're right. LGTM. Thanks

Openjpeg team just had a fair good amount of bug fixes on their side. 
Should we merge them manually or wait for the infra team'a advice on using DEPS?

[Bo Xu, 2014-10-21 19:17:41]

Committed patchset #1 (id:1) manually as
bff247d92c315f9aab2edb4aeaf0ef0b74ba321f (presubmit successful).

[Tom Sepez, 2014-10-21 19:18:00]

> Openjpeg team just had a fair good amount of bug fixes on their side. 
> Should we merge them manually or wait for the infra team'a advice on using
DEPS?
For now, you're gonna have to merge them manually.  
I haven't got a response from those guys despite pinging.