胡斌 / merge_av

1.没有老师，但是有两个学生时，默认使用一对多布局
2.一对一时，老师和学生的视频尺寸不同时，合成的视频高度取老师视频高度加上学生视频高度（之前版本假定老师和学生视频大小一致，使用的是最大高度的二倍）
3.一对多时，支持学生的视频尺寸不一致的情况（之前版本假定只有一种视频尺寸，其它尺寸视为异常）
4.一对多时，最多只在老师视频下部排4个学生，多于4个学生的情况，扩大合成视频的高度，把学生视频排列在老师视频下方，保证老师视频部分最多只有一排学生视频，避免老师视频被过度遮挡

it is to be improved in one to many case if the size of each student frame is different

2.get the max width and height of teacher and student,to deal with case that the width or height is not qual of student and teacher,not finished

1.修改老师和学生视频大小不一致，老师的视频小于320x240时，老师背景有花屏
2.修改老师和学生视频都为240x180视频时，视频宽度仍为240, 且有花屏
3.只有一路视频时，默认合成为上下布局

but at present,only support landscape layout

parse width and height in record info file

adjust the layout of one to many layout

improved adding new file when start time reached

print and return error if no paramter after "-t"

for 0,layout is decided by audio detect
for 1,using one2one layout
for 2,using one2many layout

2.flip image when retote 180
3.save record info as m_*
4.use stream time base when decode,don't use decode codec time base,because it is varied in some mp4 file

fix the bug for RGB2YUV

save output info

because the time offset is not begin from the recordinfo file,but maybe from the last recordinfo file with same uid.

2.add macro to support using ffmpeg codec version 58,which need pcm format using AV_SAMPLE_FMT_FLTP when encoding aac

one2one transcode works,need add still frame when one parter has no video but has audio
add some code about swscale,unfinished

fix bug for finished transcoding without all files transcoded