Add receiver session timeout API callback and command line option for exit on session timeout

See merge request rist/librist!265

Add peers figures to receivers statistics.

See merge request rist/librist!263

Fixed a memory leak caused by authentication failures.

See merge request rist/librist!262

Add config file support for ristsender and ristreceiver.

See merge request rist/librist!264

The file needs to be in yaml format, here is an example:
---
secret: blarg
buffer: 0
encryption-type: 256
profile: <simple/main/advanced>
stats: 1000
inputurl:
  - <URL A>
  - <URL B>
  - ...
outputurl:
  - <URL A>
  - <URL B>
  - ...
We used a built-in very simplistic native yaml parser and not a yaml library to not introduce a new dependency. The key-value pairs are the same ones used in the long-options of the command line (all are supported).

Flow timeout

See merge request rist/librist!261

Merge latest dev branch

Closes #178

See merge request rist/librist!260

Enhance test app to have a random number of null segments and also check for content accuracy on every segment

Refactor null packet deletion to be fully compliant with the spec. No more assumptions of where the nulls can be.

Fix bug in sender rtt prometheus stats. They should be displayed in units of second and not millisecond
Closes #178

Force authenticator to use derived key when no encryption passphrase is provided.

See merge request rist/librist!259

Fix new NPD test: we cannot repeat port numbers or the tests interfere with each other as they are executed in parallel

This commit fixes an encryption issue preventing the communications:
when no encryption passphrase is provided, the authenticator requests
the authenticatee to use a derived password but does not apply this recommendation
to itself as it should.

Null packet deletion fix: do not assume we have 7 mpegts segments per packet during null restoration.

Enhance test application: 1- Use random packet sizes. 2- Emulate mpegts and check sync bytes. 3- Add tests for null packet deletion mode.

Fix authentication issue in release builds.

See merge request !258

Fix building srp_examples.c with musl

See merge request !257

assert macros prevent the code from being executed when compiling for
release targets.

Fix random data corruption on the rist receiver upon arrival of keepalive packets. The code was incorrectly using memcpy to populate an unpacked structure.

This fixes the immediate build failure, but it's quite a fragile fix,
because if any other libc defines malloc, calloc, or free in a header
that has not been included by this point in srp_examples.c, but is
subsequently included by srp.c or srp_constants.c, the same problem
will occur.  The ideal thing to do would be to find a way to
accomplish the desired effect without needing to use a macro to
override the definitions of libc functions.

We were using payload_type < 200 and now we use a range outside 200 to 205 for the classification.
When the marker bit of the rtp header is turned on, the payload type of the rtp packet could match the range we were looking for, thus incorrectly classifying rtp data as rtcp data. For example, STMPE-2110 uses rtp payload types 96 through 127 (224 to 255 with the marker bit on).
With this fix, librist will also classify rtp packets with a payload type of 72-77 as rtcp. Since that range is reserved for “RTCP conflict avoidance” on the RFC4855 standard, we can leverage that to add a feature to the library (keep reading).
Lastly, let it be documented that the library WILL only classify rtp packets as rtcp if they live in the range 200-205 or 72-77. Everything else is considered rtp data.
If anyone wants to use custom rtcp packets not defined in the RIST specification and they want their implementation to work with librist, they must use the 72-76 range or they will be treated as rtp data and run the risk of corrupting their data stream.

More Readme updates

See merge request rist/librist!256

# Conflicts:
#   README.md

Readme updates

See merge request rist/librist!255