Please describe the problem:
While reading a MPTS file, there is some cases where PAT changes are not always signals (via a gst_post_message).


Steps to reproduce:
1. take 2 MPTS file stream (recorded from a dvb device)
2. Use a MPTS file which mix 2 MPTS stream. A such file can be created with dd :  dd bs=188 count=100000 if=stream1.ts of=mix.ts ; dd bs=188 count=100000 if=stream2.ts >> mix.ts ; dd bs=188 count=100000 if=stream1.ts >> mix.ts
3. run the following command: gst-launch -m filesrc location=mix.ts ! mpegtsparse | grep pat


Actual results:
There are only two pat changes reported. There should be three : one for each part of the file.
In fact the last pat change (when mpegtsparse is back to the stream1) is ignored.

Expected results:
There should be three PAT announcement.

Does this happen every time?
yes

Other information:
In fact, subtables are decoded for the PAT table which doesn't make sense.
A proposed patch inhibits the subtable mechanism when parsing the PAT.

It also clean mpegtspacketizer stream's structure when streams are out of date (on pat change).

Note: I can send a mix.ts file on demand.