Python autotests for sample apps

[maxsharabayko]

Create autotests for srt-live-transmit and srt-file-transmit.
See /scripts/python/test-apps.py integrated into .travis.yml.

1. Generate payload and feed to the sending instance of srt-live-transmit via stdin (file://con).
2. Connect two instances on a localhost via srt protocol.
3. Read out data from stdout (file://con) of the receiving srt-live-transmit.
4. Validate data received.

The pipeline:

file://con {srt-live-transmit} -> srt(localhost) -> {srt-live-transmit} -> file://con

TODO:
Travis

• Run the script for the OSX build. At the moment python 3 is not installed there. Add the dependency.
• Try to have python 3.6+ for cool string formatting features of python. Currently v3.4 on Linux configuration.
• Make the test discoverable by Travis. Currently the test is just run in the flow of a CI script. Still the result is detected.

test-apps.py

• Start the sending function in a separate thread to add a timeout for the test.
  time.sleep(0.01) * maxLoops

srt-live-transmit

• srt://:4200 -> srt://127.0.0.1:4200
• srt://127.0.0.1:4200 -> srt://:4200
• Catch connection messages on stderr. Currently does not read. Flush does not help.

srt-file-transmit

• Allow file://con target. At the moment the app will complain about the mismatching stream IDs of the sender and the receiver.
• srt://:4200 -> srt://127.0.0.1:4200
• srt://127.0.0.1:4200 -> srt://:4200

[zerodefect]

Thinking about this a bit more, is it worth generating a one-way hash (MD5?) of the 'sent' and then comparing it to the 'received' for additional validation? I think comparing file size too for added measure.