msm: mdss: fix wrong chroma stride for h2v1/h1v2 formats

Current driver is swapping the chroma subsample
for the read pipe when rotation is in place,
this configuration is already taken care by the
rotator, and it is not needed to swap the chroma
sampling for the rotator pipes.
This fixes wrong stride calculations, causing
wrong colors shown in h2v1/h1v2 tests.

Change-Id: I43a841d6030636f71f271cc21f8177b44b2c0ca3
Signed-off-by: Ingrid Gallardo <ingridg@codeaurora.org>