BNPH-5393: --resolve add vars to affect mod_voicemail behavior: skip_record_check and skip_record_urgent_check

fix the vp8 issue, but the other part causes issues with centos5... we\'ll need better detection there and working on a way to handle this

Add record_seconds, record_ms and record_samples variables done recording.
original patch not used due to refactor in code.

Bad things must have been happening when this was hit.

This was introduced in commit 5cb4cd9d

FS-5148 gsmopen: wait for the first audio write() before giving the AT command that moves the audio path to the audio tty (eg: don't accumulate noise by not sending frames on the open tty)

gsmopen: on Linux, automatically find out and setup audio and data ports associated to dongle model, IMEI and IMSI

FS-5568 --resolve the real problem is that L=-1 is not valid because the L param is about how many times to generate the output while generating the tone so -1 is impossible because it can never end to let you hear it.  The real fix is to not allow -1 loops=-1 is the correct way to do this because that is parsed after the tone is generated and repeats the entire thing.  Doing something like L=100 actually generates the whole slinear audio 100 times into memore where loops=100 only generates the little snippet and repeats it after it was generated.