Reduce number svn blame calls

[Peter Johansson]

related to #358

svn blame calls are very expensive; try with svn-client.

It depends, of course, on the connection to the server how large fraction of the run time that we can blame on SVNblame, but I think we can reduce the total run time significantly by minimizing number of calls to SVNblame.

There are two places to reduce number of svn blame:

1. obsolete The first place is in the output (as described in ticket:358). If we wanna keep the output as now, we should at least move the svn blame call so it is only done once for each File and then pass around the SVNblame object to the several outputs.

2. In the parsing, one svn blame is called for every rev the File has been modified. However, this happens twice since it happens for both SVNblame and SVNadd. If we could re-structure the code, so only one svn blame call is needed and that information can be passed to both classes.

[Peter Johansson]

spelling

[Peter Johansson]

The loop over the different outputs for one page occur in Node, and therefore it would be a bit complicated to implement 1). As ticket:358 include the same reduction of SVNblame call and it seems to be easier to implement, I'll go for ticket:358 instead.

2) might still be an idea to consider, so I leave this ticket open.