[mythtv] More scheduling scheduler

[Max Barry]

I see the softpad branch has been updated! Here's the results of a
reasonably exhaustive test of rev 11301:

SUMMARY
-------
Worked well until the schedule got complicated, then broke badly.

CONDITIONS
----------
* start-early softpad of 1 minute
* end-late softpad of 2 minutes
* no pre-/post-roll
* no individual hard start-early or soft-late
* 2 tuner cards (IDs 3 and 4)
* no tuner/input priority

WHAT WORKED
-----------
The basics: programs were padded with the extra minutes if possible, and
moved to idle tuners if necessary. This soft padding was dropped where
necessary to avoid causing conflicts.

The softpad was conveniently and accurately displayed in "(+1 +2)"
format in Upcoming Recordings, and programs with missing softpad were
slightly darkened. This was very helpful in spotting sub-optimal schedules.

Not a real problem, but I will mention that programs now display like
"Single Record +0/+3 Will Record". That "+3" is new and cryptic (to me).

WHAT DIDN'T
-----------
The more programs you try to cram in, the less reliable softpad gets. At
fairly low levels of complexity (e.g. 2 tuners and 3 programs), soft
padding can be dropped when it doesn't need to be. See attachment
softpadtest1.txt for a detailed example.

More seriously, I was able to get MythTV to conflict a program solely
because of softpad--which should never happen. See softpadtest2.txt for
details. This required me to schedule 5 programs on 2 channels with 2
tuners, a situation of unusually high load, but it's not unthinkable,
either.

Max.

-------------- next part --------------
An embedded and charset-unspecified text was scrubbed...
Name: softpadtest1.txt
Url: http://mythtv.org/pipermail/mythtv-dev/attachments/20060927/5020abd3/attachment.txt 
-------------- next part --------------
An embedded and charset-unspecified text was scrubbed...
Name: softpadtest2.txt
Url: http://mythtv.org/pipermail/mythtv-dev/attachments/20060927/5020abd3/attachment-0001.txt