If you want to take a few minutes and read the lengthy introduction to this question, it might give you a better idea of why I’m asking it.  If not, just go to the last paragraph.
 
I vividly recall an instance from a Junior High English class where we all had to give an oral report on any topic that was of interest to us.  One classmate gave his report on ice hockey.  He explained how the game was played and gave an extensive glossary on the terminology of the game.  Time after time, he would state the next hockey term followed by, “…which is self-explanatory”.  Then he mentioned the term, “checking”, followed by, “which is self-explanatory”.  To anyone not familiar with the game, slamming your opponent into the side boards has nothing to do with where you keep your money in order to write checks.  We can become so accustomed to a specific way of speaking about any subject that others may not have any idea of what we are trying to say.
 
Ah, you must be thinking that I’m a Newbie to recording.  Not so.  I had my share of bouncing multiple tracks down to one track in the good old reel-to-reel tape days.  But it does not appear that this is what’s happening here. There is no limit to the number of tracks available in Sonar, so the old school method of bouncing tracks is meaningless.
 
I have watched several video demonstrations of the Sonar bounce function and have yet to have an explanation of why it’s being done.  All they say is, “Okay, let’s bounce this clip”.  Huh?  Where are you bouncing it?  Can somebody explain to me what is taking place?  In other words, #1) why do it, and #2) what is being accomplished.  (I’ll be checking your answers).

Bounce to clips changes the way the data is packaged, allowing operations not possible with 'raw audio'. Packaged up, you can drag repeated copies out to extend the audio w one mouse operation, for example.
 
Bouncing to tracks is similar to what you did on MTR tape, selecting one or more tracks and combining them on a new track. Mixdown. Note that you can also 'mixdown' a single track by bouncing, so as to create a new audio track with FX and automation included in the audio data, to free up processing power. Once 'mixed', it is much more difficult to 'extract' the original data to process it differently. Term here is 'destructive editing'.
 
This is a simple explanation, it gets deeper from here.

In fact Sonar is a non destructive editor. Another word for bounce to, is render.
With Sonar tools you perform edits on audio or midi data or process midi or audio data through plugins. When you process the edited or effected data into an audio file it's called bouncing. While Sonar processes the data it creates a new audio file leaving the pre edited or processed data in tact.
The Sonar.cwp file is where all the MIDI and edit data is kept. The audio files are kept separate in a folder connected to the project.
There are some process fx which are destructive but are accessed through a special menu with check ks that make sure you understand the permanence of your intended action.

If you have used the scissors tool to split clips, the only way I know to put them back together—once it's too late for UNDO—is to highlight them and then select Bounce to Clips.
 
Example: One piece of a vocal recording is a little late.  I split the clip so that section is a separate clip.  I nudge the clip to the left until is sounds in time.  Then I highlight the clip before it and after it and select Bounce to Clips to put it all back together:  NOTE: There is no reason you have to put i back together.  I do because I like all the various lead and harmony vocals to start and end at the same time so I can move them around.  If Chorus 1 starts at 16:4:3:60 then I copy it to 32:4:3:60 and I know it will be right.  It the doubled lead, harmonies, etc. have all been trimmed to start at the same time, it is way easier to copy and move stuff.  Plus it looks neater.
 
Related trick: you have a vocal singing chorus 1 and chorus 2.  Clone the track.  On the clone, reverse chorus 1 and chorus 2 and you have instant doubling.  I do this with harmonies all the time.  Easier if the times are consistent.

konradh
If you have used the scissors tool to split clips, the only way I know to put them back together—once it's too late for UNDO—is to highlight them and then select Bounce to Clips.
 
Example: One piece of a vocal recording is a little late.  I split the clip so that section is a separate clip.  I nudge the clip to the left until is sounds in time.  Then I highlight the clip before it and after it and select Bounce to Clips to put it all back together:  NOTE: There is no reason you have to put i back together.  I do because I like all the various lead and harmony vocals to start and end at the same time so I can move them around.  If Chorus 1 starts at 16:4:3:60 then I copy it to 32:4:3:60 and I know it will be right.  It the doubled lead, harmonies, etc. have all been trimmed to start at the same time, it is way easier to copy and move stuff.  Plus it looks neater.
 
Related trick: you have a vocal singing chorus 1 and chorus 2.  Clone the track.  On the clone, reverse chorus 1 and chorus 2 and you have instant doubling.  I do this with harmonies all the time.  Easier if the times are consistent.

I do this quite a bit, but always start by sorting the timing inconsistencies out in V-Vocal, then bounce V-Vocal afterwards, then clone & swap