avcodec/mips: [loongson] refine ff_vc1_inv_trans_8x8_mmi.

Combined 1st and 2nd loop into one inline asm in function ff_vc1_inv_trans_8x8_mmi to
reduce memory operation, and made some small optimization in ff_vc1_inv_trans_4x8_mmi.

Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>